....
71h.
. , nu .-
s. , ‘ , ‘ , m."
19 .. n a ‘ , . . Eu...
91;: fan..." . (u
3.. 5:: .
.. .1

.z r... v .
5.15%.“. .
.21 r 3..
E . . .r t

urn.

, . .. ‘4
, 3...! z Li
. . 5.35.. 33.5..
. . . 971;?! .. ten: : bl
, , . 3323.27.13.
. . an .14} ’ u. t
2.5.
. ix} .
.2". . . a r

.5. $1.. I. .

in: a. .11).! ‘ u 1! a:

a I I.
1...}.

. we.
a . .
Inuit...)
i':i. I

we...

:7

. a
.:. z . .
}....N.:.;
3:"... nfifiaflh
.39.. . r
a... 1 e. .
Iaé
3.3.5.3.
5.5...

ummwfié

i a

5.1;... fish“; ..

7“: u {I‘lf-u .
nu. 5A.}: a [2. . f
‘ is. that‘s: an
2.l3"fliy§b.i (“naoblkp
. . 'QI
’{g’im‘vn
.tl 1|Jt01.f:.vltirl
I}!!! [iii-$112...
u. i .0 SI: (4-
5.1533}.
.3 .i!
:n
1...... ‘ ..

a march...

alt-Erlii‘éilt {I
1‘. 3...!!! it... 7
Bh}'.tih.ftlfi$lg

: 31312.53 9

.. . 1%.... :l.
3 la’ili..xlr!.i§ 3v.
..£\.1 .4 «15:31.1...

’31:- I\ I: I4.»

3. Lu ~11. 5.?

[art‘s-.4?

b .i

.1 t.l(d..l.(14.‘1al.).llx
‘ allotr‘égvsrx.
x 1 I.

 

bv'v

”A,

.9. Lt I

 

LIBRARY

Michigan State
University

 

 

 

This is to certify that the

dissertation entitled

VARIABLE FREQUENCY MICROWAVE PROCESSING AND MICROWAVE
PROCESS CONTROL FOR POLYMER COMPOSITES

presented by

Yunchang Qiu

has been accepted towards fulfillment
of the requirements for

Ph.D. Chemical Engineering

degree in

 

 

WW~

I Major professor

Date Aer/Mg?”

MSU is an Affirmative Action/Equal Opportunity Institution 0-12771

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

11/00 chIRCIDmDuMJA

 

VARIABLE FREQUENCY MICROWAVE PROCESSING AND MICROWAVE

PROCESS CONTROL FOR POLYMER COMPOSITES

By

Yunchang Qiu

A DISSERTATION
Submitted to
Michigan State University

in partial fiilfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Chemical Engineering

2000

ABSTRACT
VARIABLE FREQUENCY MICROWAVE PROCESSING AND MICROWAVE
PROCESS CONTROL FOR POLYMER COMPOSITES
By

Yunchang Qiu

This dissertation presents the research work on the development of a variable
frequency microwave processing system for polymer and composite materials, with an
emphasis on achieving uniform temperature distribution using intelligent process control.
A variable frequency microwave material processing system was constructed based on the
existing fixed fi'equency microwave processing technology with the use of a variable-
frequency microwave power source. Data acquisition and control hardware was
implemented for process monitoring, measurement, and control. Software programs were
developed in LabVIEW for data acquisition, system characterization, and process control.
The control objective is to achieve efficient, uniform, and controlled heating, which was
realized by mode tuning, intelligent mode switching, on-line mode characterization, and
effective power control.

Two uniform processing techniques were developed and evaluated. They are mode
sweeping heating, and intelligent mode switching heating. Mode sweeping heating proved
to be very effective for small size samples. Intelligent mode switching heating Optimizes
the sequence of the modes used for heating, by comparing the mode heating
characteristics with measured temperature distributions and selecting the mode that will

alleviate the temperature gradients the most. Using intelligent mode switching heating,

great improvement of temperature uniformity was achieved over single mode heating and
mode sweeping heating. An on-line mode characterization technique was developed to
enable the process control system to adjust to process condition changes. With the
addition of on-line mode characterization capability, consistent and good performance was
ensured for the variable frequency microwave processing system, as demonstrated by the
uniform and stable processing of composite parts with complex geometry.

During the mode selection process in the intelligent mode switching heating,
modes were compared by their ability to decrease the temperature gradients and generate
the most uniform temperature distribution within a desired period of time. Temperature
uniformity was measured by the standard deviation of the temperatures. Therefore, the
optimal mode would decrease the temperature standard deviation the most within the
specified period of time. Two power control algorithms were designed to achieve the
objectives of providing fast heating, reducing temperature overshoot, and maintaining
constant curing temperature. Consequently, a simple parabolic power controller and a
multi-staged PID controller were designed for microwave power control. The former
needed no controller tuning, while the latter required proper tuning of the control
parameters but provided more stable and accurate control performance.

The experimental results proved the variable frequency microwave processing
system successful in achieving uniform processing with consistent performance. A Viable
variable frequency microwave processing system for industrial applications can be
developed based on this newly developed system. The full automation of the hardware and
the flexibility of the software ensure easy implementation and make the system adaptable

to different types of applications.

Copyright by
Yunchang Qiu

2000

To my wife Fang,
my parents, ZhaoLong and FuZhao,
and my brothers, Bao, Yu, and Hong,

for the unconditional love and everlasting inspiration.

ACKNOWLEDGMENTS

I wish to thank Dr. Martin C. Hawley for his invaluable guidance and tremendous
support throughout the research work and editorial process for this dissertation. Thanks
are also extended to Professor Jes Asmussen, Professor Kun-mu Chen, Professor Larry
Drzal, Dr. Jianghua Wei, and Dr. Valerie Adegbite for many enlightening discussions and

their insightfiil advice and suggestions.

This research was funded by the NSF I/UCRC Polymer Processing Center at

Michigan State University.

vi

TABLE OF CONTENTS

LIST OF TABLES .......................................................................................................... xi
LIST OF FIGURES ....................................................................................................... xii
INTRODUCTION .......................................................................................................... 1
CHAPTER 1
MICROWAVE PROCESSING FUNDAMENTALS AND LITERATURE REVIEW... 13
1.1 Electromagnetic Theory ..................................................................................... 13
1.1.1 Electric Field and Magnetic Field ............................................................... 13
1.1.2 Fundamental Electromagnetic Theory ........................................................ 17
1.1.3 Boundary Conditions ................................................................................. 20
1.1.4 Electromagnetic Fields in a Cylindrical Cavity ............................................ 22
1.1.4.1 TE Modes .............................................................................................. 23
1.1.4.2 TM Modes ............................................................................................. 24
1.1.4.3 Mode Designation .................................................................................. 25
1.1.4.4 Electric Field Pattern .............................................................................. 25
1.1.4.5 Cut-off Frequency .................................................................................. 28
1.1.4.6 Cavity Quality Factor ............................................................................. 30
1.2 Interactions Between Microwaves and Materials ................................................ 31
1.3 Microwave Processing of Materials .................................................................... 36
1.3.1 Microwave Processing of Polymers ............................................................ 36
1.3.2 Microwave Processing of Composites ........................................................ 37
1.3.3 Kinetics of Microwave Curing of Epoxies and Epoxy Composites .............. 38
1.3.4 Other Microwave Heating Applications. . ............................................ 43
1.4 Variable Frequency Microwave Material Processing .......................................... 43
1.5 Process Modeling and Processing Control in Microwave Processing .................. 45
1.5.1 Microwave Process Modeling .................................................................... 45
1.5.2 Microwave Process Control ....................................................................... 49
CHAPTER 2
VARIABLE FREQUENCY MICROWAVE PROCESSING SYSTEM AND
COMPUTER CONTROL INSTRUMENTATION ....................................................... 52
2.1.1 Variable Frequency Microwave Power Source ........................................... 53
2.1.2 Cylindrical Single-Mode Resonant Cavity ................................................... 53
2.1.3 Other Microwave Circuit Components ....................................................... 55
2.2 Automation of the Variable Frequency Microwave Power Source ...................... 55
2.3 Computer Data Acquisition and Control Implementation ................................... 55
2.3.1 Measurement Instrumentation .................................................................... 56
2.3.1.1 Temperature Measurement ..................................................................... 56
2.3.1.2 Power Measurement ............................................................................... 56
2.3.2 Control Instrumentation ............................................................................. 57

vii

2.3.2 1 Frequency Control ................................................................................. 57

2.3.2.2 Power Control ....................................................................................... 57
2.3.3 Computer Data Acquisition ........................................................................ 58
CHAPTER 3
CHARACTERIZATION OF VARIABLE FREQUENCY MICROWAVE
PROCESSING SYSTEM ............................................................................................. 59
3.1 Variable Frequency Method ............................................................................... 59
3.2 Variable Frequency Microwave Power Source Characteristics ........................... 61
3.3 Characterization of the Empty Cavity ................................................................. 64
3.4 Characterization of the Loaded Microwave Cavity ............................................. 65
3.5 Variable Attenuator Characteristics .................................................................... 69
3.6 Power Meter Response Time ............................................................................. 71
3.7 Frequency Effects on Material Properties ........................................................... 74
3.7.1 Dielectric Measurements of Uncured DGEBA/DDS ................................... 74
CHAPTER 4
VARIABLE FREQUENCY MODE SWEEPING HEATING ....................................... 77
4.1 Experimental Preparation ................................................................................... 77
4.2 Mode Heating Characteristics ............................................................................ 79
4.3 Complementary Heating Concept and Mode Switching Technique ..................... 84
4.4 Control Algorithm and Program ......................................................................... 86
4.5 Selective Mode Sweeping Heating Results ......................................................... 86
4.6 Summary and Conclusions ................................................................................. 89
CHAPTER 5
INTELLIGENT VARIABLE FREQUENCY MODE SWITCHING PROCESSING ..... 90
5.1 Rationale of Intelligent Mode Switching Heating ............................................... 90
5.2 Process Control System - VFMPCS I ................................................................ 92
5.2.1 Mode Tuning Controller ............................................................................ 93
5.2.2 Mode Selection Algorithm ......................................................................... 94
5.2.3 Power Control Algorithm ........ . .................................................................. 95
5.3 Variable Frequency Microwave Processing of Square Graphite/Epoxy Composite
Parts .................................................................................................................. 97
5.3.1 Microwave Power Adjustment Using Stepper Motor .................................. 97
5.3.2 Experimental Results and Discussion .......................................................... 97
5.4 Variable Frequency Microwave Processing of V-shaped Graphite/Epoxy
Composite Parts ............................................................................................... 104
5.4.1 Microwave Power Adjustment Using Variable Attenuator ......................... 104
5.4.2 Experimental Results and Discussion ......................................................... 106
5.5 Summary and Conclusions ............................................................ . ................... 116
CHAPTER 6
VARIABLE FREQUENCY MICROWAVE PROCESSING OF COMPLEX SHAPE
COMPOSITE PARTS WITH ON-LINE MODE UPDATING ..................................... 118

viii

6.1 On-line Updating of Mode Heating Characteristics ............................................ 118

6.2 Process Control System - VFMPCS II .............................................................. 119
6.2.1 Mode Tuning Controller ........................................................................... 119
6.2.2 Mode Selection Controller ........................................................................ 120
6.2.3 Multi-staged PID Microwave Power Controller ........................................ 121
6.2.4 On-line Mode Characteristics Updating Controller .................................... 123

6.3 Variable Frequency Microwave Processing of V-shaped Graphite/Epoxy

Composite Parts with On-line Mode Updating .................................................. 124
6.3.1 Heating Modes and Their Characteristics .................................................. 124
6.3.2 Mode Switching Heating Results and Discussion ....................................... 124
6.4 Variable Frequency Microwave Processing of Tri-planar Graphite/Epoxy
Composites with On-line Mode Updating .......................................................... 131
6.4.1 Heating Modes and Heating Characteristics ............................................... 132
6.4.2 Intelligent Mode Switching Heating Results and Discussion ...................... 138
6.5 Summary and Conclusions ................................................................................ 144
CHAPTER 7
SUMMARY AND CONCLUSIONS ........................................................................... 145
7.1 Development of an Automated Variable Frequency Microwave Processing
System .............................................................................................................. 146
7.1.1 Automation of the Microwave Processing System ..................................... 146
7.1.2 Characterization of the Variable Frequency Microwave Processing System147

7.2 Variable Frequency Mode Sweeping Heating .................................................... 148

7.3 Variable Frequency Mode Switching Processing ............................................... 149

7.4 Variable Frequency Microwave Processing of Complex Shape Composite Parts

with On-line Mode Updating ............................................................................ 150

7.5 Summary .......................................................................................................... 152

CHAPTER 8
RECOMMENDATIONS AND FUTURE WORK ........................................................ 156
8.1 On-line Cure Monitoring for Microwave Processing of Polymers and
Composites ....................................................................................................... 1 57
8.2 Hybrid Heating for Ultimately Uniform Processing ............................................ 158
8.3 Scale-up Studies and Industrial Application of Variable Frequency Microwave
Processing System ............................................................................................ 160
APPENDICES ............................................................................................................. 161
APPENDIX A
Control Hardware Instrumentation ............................................................................... 162
APPENDIX B
LabVIEW Subvi's ............................... . ......................................................................... 175
APPENDIX C
LabVIEW Programs for System Characterization ......................................................... 185
APPENDIX D
LabVIEW Programs for Process Control System .......................................................... 196

ix

BIBLIOGRAPHY ........................................................................................................ 213

LIST OF TABLES
Table 1.1 Boundary Conditions .................................................................................... 21
Table 3.1 Time for the Computer to Write Frequency to the Oscillator ......................... 63
Table 3.2 Experimental Measurement of Resonant Modes in an Empty Cavity .............. 64
Table 3.3 Power Difference Percentage afier l70ms. .................................................... 73
Table 3.4 Dielectric Properties of Uncured DGEBA/DDS ............................................ 75
Table 4.1 Frequency Shift of Empirical Modes Due to Temperature Change ............... . 80
Table 5.1 Frequencies of the Modes Used in the Mode Switching Heating ................... 102

Table 6.1 Average Maximum Temperature and Standard Deviation at Curing Stage 126
Table 6.2 Average Maximum Temperature and Standard Deviation at Curing Stage 140
Table A.1 Connector and cable wire assignments ......................................................... 162

Table A.2 I/O Connector Block Temrinals and Corresponding Signals ......................... 168

LIST OF FIGURES

Some images in this dissertation are presented in color.

Figure 1.1 Electric Field Patterns for TE Modes ........................................................... 26
Figure 1.2 Electric Field Patterns for TM Modes .......................................................... 27
Figure 1.3 Mode Chart for a 7 Inch Diameter Empty Cavity ......................................... 29
Figure 1.4 Q Factor Calculation using Half Power Point Method .................................. 31
Figure 1.5 Dissipation Factor 6' vs Temperature for DDS, DDM and DDE ................. 41
Figure 2.1 Variable Frequency Microwave Processing System ...................................... 54
Figure 2.2 Cylindrical Single-Mode Resonant Cavity .................................................... 55
Figure 3.1 Mode Chart of the Empty Cavity ................................................................. 60
Figure 3.2 Microwave Power Source Output Versus Frequency Curve ......................... 62
Figure 3.3 Comparison of Theoretical and Experimental TM012 Mode Curves ............. 65
Figure 3.4 Power Reflectance versus Frequency During Frequency Scan ...................... 66
Figure 3.5 Temperature Change versus Frequency During Frequency Scan ................... 67
Figure 3.6 Relationship between Control Voltage and Power Attenuation ..................... 70
Figure 3.7 Measured Microwave Power versus Time after Power Step Change ............ 72
Figure 3.8 Difi‘erence Percentage versus Time after Power Step Change ....................... 72
Figure 3.9 Dielectric Constant versus Frequency for DGEBA/DDS .............................. 75
Figure 3.10 Dielectric Loss Factor versus Frequency for DGEBA/DDS ........................ 76
Figure 3.11 Q-Factor versus Frequency ........................................................................ 76
Figure 4.1 Composite Material Lay-up Procedure ......................................................... 78
Figure 4.2 Schematic Sketch of the Teflon Mold .......................................................... 79

xii

Figure 4.3 Temperature Measurement Locations .......................................................... 79

Figure 4.4 Thermal Paper Images of Four Selected Modes ........................................... 81
Figure 4.5 Single Mode Heating Temperature Profiles of Selected 4 modes .................. 83
Figure 4.6 Complementary Heating Using Two Modes ................................................. 84
Figure 4.7 Mode Sweeping Algorithm .......................................................................... 85
Figure 4.8 Thermal Paper Images of Mode Sweeping Heating ...................................... 87
Figure 4.9 Mode Sweeping Heating Temperature Profiles ............................................ 88

Figure 5.1 Process Control Diagram for Variable Frequency Mode Switching Heating . 92

Figure 5.2 Power Temperature Relationship ................................................................. 96
Figure 5.3 Temperature Measurement Locations .......................................................... 98
Figure 5.4 Percentage of Reflected Power versus Frequency ......................................... 98
Figure 5.5 Single Mode Heating at f=2.5737 GHz ....................................................... 100
Figure 5.6 Single Mode Heating of 3" Square Sample - f = 3.0818 GHz ...................... 101
Figure 5.7 Mode Switching Heating of 3" Square Sample ............................................ 103
Figure 5.8 Mode Selection Histogram during Mode Switching Heating ....................... 103
Figure 5.9 Input Power Change during Mode Switching Heating ................................. 104

Figure 5.10 Comparison of Power Control Performance between Variable Attenuator and

Stepper Motor ...................................................................................................... 105
Figure 5.11 V-shaped Sample and Teflon Mold Configurations ................................... 106
Figure 5.12 Temperature Measurement Configuration ................................................. 107
Figure 5.13 Power Reflectance versus Frequency ........................................................ 107
Figure 5.14 Temperature Change during Frequency Scan ............................................ 108
Figure 5.15 Temperature Profiles of Single Mode Heating ........................................... 112

xiii

Figure 5.16 Single Mode Heating at f = 3.6506 GHz ................................................... 113
Figure 5.17 Mode Switching Heating 1 with Curing Temperature Control Window:
155°C - 160°C ...................................................................................................... 114

Figure 5.18 Mode Switching Heating 2 with Curing Temperature Control Window:

157°C -160°C ...................................................................................................... 116
Figure 6.1 Multi-Staged PID Control .......................................................................... 122
Figure 6.2 Temperature Measurement Configuration ................................................... 124

Figure 6.3 Intelligent Variable Frequency Mode Switching Heating of V-shape
Graphite/Epoxy Composite with On-line Mode Updating ...................................... 128
Figure 6.4 Mode Sweeping Heating of V-shaped Graphite/Epoxy Composite .............. 129
Figure 6.5 Single Mode Heating at f = 2.1605 GHz for V-shaped
Graphite/Epoxy Composite ................................................................................... 130
Figure 6.6 Comparison of Temperture Uniformity for Single Mode Heating, Mode

Sweeping, and Intelligent Mode Switching Heating of V-shaped Graphite/Epoxy..131

Figure 6.7 Configuration of Tri-planar Graphite/Epoxy Samples .................................. 132
Figure 6.8 Temperature Measurement Configuration of Tri-planar Samples ................. 132
Figure 6.9 Power Reflectance versus Frequency Curve for a Tri-planar Sample ........... 133
Figure 6.10 Temperature Change during Frequency Scan ............................................ 134
Figure 6.11 Mode Heating Characteristics ................................................................... 137
Figure 6.12 Single Mode Heating Profile at f = 3.8326 GHz ........................................ 138
Figure 6.13 Intelligent Mode Switching Heating of Tri-planar Graphite/Epoxy ............ 141
Figure 6.14 Mode Sweeping Heating of Tri-planar Graphite/Epoxy ............................. 142

xiv

Figure 6.15 Temperature Uniformity Comparison of Single Mode heating, Mode
Sweeping Heating, and Intelligent Mode Switching Heating of Tri-planar
Graphite/Epoxy .................................................................................................... 143

Figure 8.1 An Example of Relating Power Absorption Curve Change to Extent of Cure

Change ................................................................................................................. 159
Figure A.1 Device Configuration of the Variable Attenuator ........................................ 163
Figure A.2 Pin Assignments for PCI-MIO-16XE-50 Board ......................................... 167
Figure A.3 Schematic for V-Shaped Teflon Mold Cover .............................................. 170
Figure A.4 Schematic for V-Shaped Teflon Mold Holder ............................................ 170
Figure A.5 Schematic for V-Shaped Teflon Mold Latches and Probing Holes .............. 171
Figure A.6 Schematic for Tri-Planar Teflon Mold Cover ............................................. 172
Figure A.7 Schematic for Tri-Planar Teflon Mold Holder ............................................ 173
Figure A.8 Schematic for Tri-Planar Teflon Mold Latches and Probing Holes .............. 174
Figure B.l LabVIEW Program of f-write#.vi - Front Panel and Diagram ..................... 177
Figure B.2 LabVIEW Program of valstep.vi - Front Panel and Diagram ...................... 178
Figure B.3 LabVIEW Program of vapwrctrl.vi - Front Panel and Diagram ................... 179
Figure B.4 Additional Elements of vapwrctrl.vi - Diagram ........................................... 180
Figure B.5 LabVIEW Pogram of pwrctrl.vi - Front Panel and Diagram ........................ 181
Figure B.6 Additional Elements of pwrctrl.vi Diagram ................................................. 182
Figure B.7 LabVIEW Program for m-tuning.vi - Front Panel and Diagram ............. - ' ..... 183
Figure B.8 Additional Elements of m-tuning.vi Diagram .............................................. 184
Figure C.l LabVIEW Program of vapwrtest.vi - Front Panel and Diagram .................. 187
Figure C.2 Additional Elements of vapwrtest.vi Diagram ............................................. 188

Figure C.3 LabVIEW Program of p-response-test.vi - Front Panel .............................. 189
Figure C.4 LabVIEW Program of p-reponse-test.vi - Diagram .................................... 189
Figure C.5 Additional Elements of p-reponse—test.vi Diagram ...................................... 190
Figure C.6 LabVIEW Program of characterization&temp.vi — Front Panel (Left Halt) .191

Figure C.7 LabVIEW Program of characterization&temp.vi - Front Panel (Right Half)192

Figure C.8 LabVIEW Program of characterization&temp.vi - Diagram ....................... 193
Figure C.9 Additional Elements of characterization&temp.vi Diagram ......................... 194
Figure C.9 (Continued) ............................................................................................... 195
Figure D.l LabVIEW Program of singlemode.vi - Front Panel ..................................... 199
Figure D.2 LabVIEW Program of singlemode.vi - Diagram ......................................... 200
Figure D.3 Additional Elements of singlemode.vi Diagram .......................................... 201
Figure D.4 LabVIEW Program of modesweep.vi - Front Panel .................................... 202
Figure D.5 LabVIEW Program of modesweep.vi - Diagram ........................................ 203
Figure D.6 Additional Elements of modesweep.vi Diagram .......................................... 204
Figure D.7 LabVIEW Program of VFMPCSI.vi - Front Panel ..................................... 205
Figure D.8 LabVIEW Program of VFMPCSI.vi - Diagram .......................................... 206
Figure D.9 Additional Elements of VFMPCSI.vi Diagram ........................................... 207
Figure D.9 (continued) ................................................................................................ 208
Figure D.10 LabVIEW Program of VFMPCSII.vi - Front Panel .................................. 209
Figure D.11 LabVIEW Program of VFMPCSHyi - Diagram ....................................... 210
Figure D.12 Additional Elements of VFMPCSIIvi Diagram ........................................ 211
Figure D.12 (continued) .............................................................................................. 212

INTRODUCTION

Microwaves are electromagnetic waves in the frequency range from 300 MHz to
300 GHz. Since the discovery of electromagnetic waves, it has been widely used in
communications. The earliest reported commercial use of microwaves in polymer
processing was in 1940 in an attempt to cure plywood cement [1]. Over the decades,
microwaves have been applied in polymer and composite materials processing, adhesive
and repair, ceramic materials processing, food processing, wood drying, waste treatment,
and in medical use as well. In the 19605, microwave processing was successfully applied
in the vulcanization of the rubber in the tire industry [2]. By now, the vulcanization of
extruded rubber weather-stripping for the automotive and construction industries has been
one of the most successful applications of microwave heating in industry [3]. Since the
mid-19803, there has been a resurgence of interest in the microwave processing of
polymers and composites [4-9].

Compared with conventional means, microwave heating has the advantages of
being volumetric, direct, selective and instantaneously controllable. Microwaves can
penetrate the material placed inside its fields. All the molecules of the material are subject
to the electromagnetic field, although the field strength decreases as it gets deeper into the
material. The interaction between materials and microwaves is direct and occurs as soon
as the electromagnetic field is established. The ability of the material to absorb microwave
energy and convert it to thermal energy depends largely on the dielectric properties of the
material. Microwave heating is, therefore, selective. Material with higher dielectric loss

factor can dissipate more electromagnetic energy into thermal energy than a material with

 

 

a lower dielectric loss factor. The amount of microwave energy absorbed by the material
also depends on the magnitude of the electric field strength. Higher field strength results in
faster heating, provided other conditions are the same. A desired temperature distribution
can be obtained if one can find ways to control the electric field as desired inside the
material. As a comparison, in conventional heating the difference between surface
temperature and inside temperature is the driving force. In a sense, the microwave heating
process can be viewed as having three degrees of freedom, while conventional heating as
having one degree of freedom.

The application of microwave heating in polymer and composite processing has
been shown very promising. Significant advantages over conventional heating have been
demonstrated. Examples are: increased polymerization rate for epoxy curing
(DGEBA/DDM) [10]; reduced drying time for pelletized polycarbonate and polypropylene
[11]; increased T3 for cured epoxy (DGEBA/DDS) [9]; enhanced fiber/matrix adhesion in
carbon composites [12], and increased mechanical strength of graphite/epoxy composite
[13]. Microwave energy also offers the potential for processing of materials that are
difficult to process by conventional thermal conduction methods, such as polymeric
materials that have poor thermal conductivity.

To utilize the heating effects of microwaves, a device termed a microwave
applicator is needed to effectively couple the microwave energy into the material to be
processed. There are three kinds of microwave applicators that are commonly used in
microwave processing of materials: single-mode, multi-mode, and waveguide applicators.
The single mode resonant applicator is designed to support only one resonant mode, while

results in highly localized heating. A mode has defined electromagnetic patterns.

 

Therefore, strong fields at desired regions can be established with single mode applicators.
In a multi—mode oven, several electromagnetic modes are randomly excited simultaneously
for a given applicator volume [14]. The features of a multi-mode applicator are such that it
is versatile in heating a wide range of materials, but it is not efficient in energy use and is
limited in heating uniformity resulting in unpredictable hot spots. A waveguide is a hollow
conducting pipe with either a rectangular or a circular cross-section. The wave inside a
waveguide is fundamentally different from that inside a multi-mode or a single-mode
applicator. The former is a travelling wave and the latter is a standing wave. Energy from
the microwave generator travels through the waveguide and is partially absorbed by the
process material. The remainder of the energy is directed to a terminating load. Travelling
wave applicators are primarily used for continuous processing of high-loss materials. Low-
loss materials require an excessively long waveguide or a slow processing speed to absorb
the necessary energy.

The temperature distribution inside the material heated by microwave energy is
dictated by the electromagnetic field distribution inside the material and the material
properties. Uneven heating results from an uneven electromagnetic field distribution,
inhomogeneous material properties, and the difference between material temperature and
ambient temperature. The common techniques to achieve uniform heating inside multi-
mode cavities include the use of a mode-stirrer and a turntable, as in home microwave
ovens, and frequency sweeping. The shortcomings of these techniques are unpredictable
temperature distribution and poor energy efficiency. For a given multi-mode applicator,
the various modes that can be excited may be known, however, the type of modes that are

excited at any time are unknown and cannot be controlled. Similarly for waveguides, the

type and number of modes that can be excited are fixed. Therefore multi-mode applicators
and waveguides are not controllable to compensate for varying material changes such as
size, shape, and especially material property changes during processing. Since most
materials have dielectric properties that change with temperature and chemical
composition, the tuning mechanism of single-mode cavities provides an advantage over
other applicators to compensate for the change. Due to its design and mechanism, single-
mode cavities are also much more efficient in energy use. Another advantage of single-
mode cavities is that the electromagnetic field distribution inside the cavity is more
predictable and process modeling with a single-mode cavity is computationally less
complex also. To achieve uniform heating inside a single-mode cavity, a mode switching
technique can be used to improve temperature uniformity. Modes with complementary
electromagnetic patterns can be excited selectively by adjusting the frequency or the cavity
volume.

Research efforts have been carried out to use the single-mode resonant cylindrical
cavity to achieve efficient, fast, and highly controllable processing for polymers and
composites. Chen and Lee [15] studied the cure of graphite/epoxy and graphite/PEEK
(polyether ether ketone) with TE112 mode at 2.45 GHz. They concluded that the coupling
of interactions between microwave energy and composites depended on the fiber
orientation and sample geometry in a complex manner. Vogel et a1. [16] demonstrated that
a 3-inch square, 24-ply graphite/epoxy composite could be processed in a single-mode
cavity with low input power. The heating rate and uniformity were dependent upon the
electromagnetic processing modes. Wei et al. showed [13] that both unidirectional and

cross-ply, thin and thick section graphite/epoxy composite materials could be successfiilly

4

processed using hybrid modes. Also using the single-mode resonant cavity, Fellows et a1.
[17] successfully processed polyimide graphite composite panels and planar and complex
shaped polyester glass composite materials using a fixed frequency mode switching
technique. Reported benefits of microwave processing of polymeric composites in a
single-mode cavity include enhanced mechanical properties, such as enhanced glass
transition temperature of the cured epoxy [13], enhanced conductive fiber/matrix adhesion
[12], faster processing times [ l8], and capability to control temperature excursions
[7][19].

In spite of the demonstrated advantages of microwave heating in composite
processing, current research has focused on laboratory-scale, exploratory efforts. Failure
to realize expected benefits from microwave processing is a result of inadequate
methodology for system integration, including system design, process control, and rapid
equipment prototyping. In many cases, the inability to provide steady temperature control
and uniform heating hindered the microwave processing systems from moving toward
production scale.

Typical microwave research at the lab scale involves intensive and cumbersome
manual operations. The microwave processing system was usually operated as an open-
loop system. Modes were selected by manually adjusting the cavity length and the
coupling probe depth. Automatic on/off control or manual rotation of dial knobs was used
to control the microwave power such that the temperature can be maintained as desired.
In most cases, computer data acquisition was not involved in the control decision making.

As a result, processing results varied for different microwave processing research groups.

 

Other problems encountered in microwave processing are temperature fluctuations,
instability of curing temperature, and large temperature gradients inside the material.

In order to realize the potential of microwave processing and develop a viable
microwave technology, work is needed to integrate microwave processing system design
with robust process control system development. The generation and transmission of
microwave energy is essentially an electronic process, an advantage that can be taken of
when designing control instrumentation. For the single-mode resonant cavity, while

controllability is one of the attractive attributes, it is yet to be fully utilized for the

- .
___‘

advancement of the technology.

In the first comprehensive effort to build a process control system for microwave
processing in single-mode cavity, Adegbite et a1. [20] automated the control of the fixed
frequency microwave power source and the adjustment of the resonant cavity. The
operation of the microwave processing system was significantly eased. Two different
control software programs were developed; one included all necessary control system
components to meet the process control objectives, and the other included only data
acquisition, hardware and interface instructions to facilitate an interface with a knowledge-
based system planner. Using a fixed frequency microwave power source, a mode
switching technique was employed to obtain uniform heating by adjusting cavity length
and coupling probe depth. Relatively uniform processing was achieved for 3-inch 24-ply
graphite/epoxy composite parts. However, the mechanical tuning of the cavity proved to
be a roadblock to more precise and consistent temperature control.

The scope of this research work is the development of a variable frequency

microwave processing system and the process control system for optimal processing

6

performance. A single-mode microwave cavity was used as the microwave applicator. The
variable frequency microwave processing system was developed based on the
configuration of the fixed frequency microwave processing system. The variable frequency
microwave power source was composed of a microwave signal generator, with a
frequency range from 1.7 GHz to 4.3 GHz, and a microwave amplifier. Microwave circuit
components were selected to be operational in the frequency range of 2 to 4 GHz. A
computer data acquisition and control system was designed and implemented. Measured
parameters included temperature and microwave power. Microwave frequency and power
are the two controlled parameters. The microwave frequency was controlled through the
GPIB interface between the computer and the microwave signal generator. Two
techniques were designed to control the microwave power. One was by electronically
adjusting the dial knob on the microwave amplifier through a stepper motor, and the other
used a voltage-controlled variable attenuator to attenuate the output of the microwave
signal generator.

Two types of uniform processing techniques were designed to attain uniform
processing temperature and the corresponding control software programs were developed.
One is variable frequency mode sweeping and the other is variable frequency intelligent
mode switching. Mode sweeping heating uses the modes in a cyclic fashion, while
intelligent mode switching heating selects the mode that is optimal for improving heating
uniformity. A mode tuning subprogram was utilized to ensure that microwave energy was
optimally coupled into the microwave cavity. An on-line mode characterization algorithm
was also designed to acquire accurate and up-to-date mode heating characteristics for

mode selection in intelligent mode switching heating. The input microwave power was the

processing variable regulated to control the processing temperature level. Both a PID
control algorithm and a parabolic equation based relational control algorithm were
designed and succeeded in maintaining a constant processing temperature and minimizing
reaction excursion. The performance of the variable frequency microwave processing
system was demonstrated and evaluated by curing simple- and complex-shaped
graphite/epoxy composite materials.

The significance of this work is in the development of a variable frequency
microwave processing technology that provides uniform and stable processing with
consistent performance and great flexibility and applicability. The advantages of using
variable frequency microwave technology have been explored and demonstrated. A
systematic processing procedure was established, including selection of sample loading
positions, location of the mode frequencies, characterization of the heating modes, and
finally computer controlled variable frequency microwave processing of the materials. A
complete set of variable frequency techniques has been created to optimize microwave
processing. The process control system that included optimal mode selection and robust
temperature control has been designed and developed. Specifically, this work made the
following contributions to the microwave material processing technology development:

1. The design and implementation of hardware and software for the automation

of a variable frequency microwave processing system to achieve fast and
precise control.

2. The development and implementation of a process control system using

innovative control methodologies, to achieve uniform and controlled heating

by mode sweeping or switching, mode tuning, and power control.

A microwave cavity characterization program that would determine the
frequencies of the heating modes and the optimal loading position of the
samples.

A predictive mode selection algorithm that would select an optimal heating
mode to alleviate the temperature gradients by matching the sample
temperature distribution with the heating characteristics of the modes.

Power control execution programs that provided fast and precise tuning of the
power control devices, stepper motor and variable attenuator.

An on-line mode heating preference characterization program that would
update the mode heating characteristics database so as to improve the
robustness of temperature uniformity control.

A variable frequency mode tuning program that provides fast and timely tuning
of the mode frequency so as to minimize reflected microwave power.

Analysis and characterization of the performance of microwave circuit
components, such as power meters, in variable frequency processing.
Automatic data acquisition for fast, reliable and convenient data collection,

tracking, and maintenance.

10. Demonstration of the ability of the variable frequency microwave processing

11.

system to provide uniform and controlled processing of complex-shaped
graphite/epoxy composite parts.
A robust procedure for variable frequency microwave processing of polymer

composites, including: optimization of sample loading position, location and

characterization of the modes, mode sweeping heating or intelligent mode
switching heating with the option of on-line mode characterization.

12. An intuitive graphical user interface for the operation and control of the

variable frequency microwave processing system.

The dissertation layout is as follows:

In chapter 1, related concepts and equations in electromagnetic theory are
discussed and wave equations are solved for an empty cylindrical single-mode resonant
cavity. The fimdamentals of the interaction between microwave and materials are
discussed, along with previous research efforts and results in microwave processing of
polymers and composites. Process modeling of microwave material processing is also
discussed to present a general picture of the microwave environment that should be
carefiilly considered during control system design.

In Chapter 2, the configuration and components of the variable frequency
microwave processing system are presented. The variable frequency microwave
processing system with a variable frequency power source was developed based on the
fixed frequency system configuration. The specifications of the power source, microwave
applicator, and other microwave circuit components are provided. The computer-based
measurement and control instrumentation for the processing system is discussed in detail.

In Chapter 3, the characterization results of the variable frequency microwave
processing system are presented and discussed. The cylindrical cavity was characterized to
make sure that it met the requirements for a single mode resonant cavity. A procedure for

Characterizing sample-loaded cavity was developed to locate empirical modes and

determine the heating characteristics of the modes. Power meters were tested for the

10

speed of response to power step changes. The characteristics of the stepper motor and the
variable attenuator were also determined. The experimental results on the effects of
frequency on dielectric properties are also presented and discussed.

In Chapter 4, a variable frequency mode sweeping heating technique and its
software program are presented. The concept of complementary heating is illustrated.
Experimental results for square graphite/epoxy composite samples are presented, which
demonstrate that the variable frequency mode sweeping technique provided uniform and
fast heating for composite parts of small size.

In Chapter 5, an intelligent variable frequency mode switching technique and the
corresponding control software (VFMPCSI) are presented. The control system including
mode tuning algorithm, mode selection algorithm, and the parabolic power control
algorithm are discussed. The performances of the variable attenuator and the stepper
motor in microwave power control were tested and compared. Intelligent variable
frequency mode switching heating results of both 3-inch square 24-ply and 3-inch V-
shaped 24-p1y graphite/epoxy composite parts are also presented. The results are
discussed and compared with those of single mode heating and mode sweeping heating.

In Chapter 6, an on-line mode characterization technique and the correspondingly
upgraded variable frequency mode switching control system (VFMPCSII) are presented.
The necessity and benefits of on-line mode characterization are discussed. A multi-staged
PID control algorithm was developed for microwave power control, the objective of
which was to provide more stable and accurate curing temperature while reducing
temperature overshoot. Experimental results of processing complexly shaped composite

parts using the process control system with on-line mode characterization (VFMPCSII)

11

are presented. The results are discussed and compared with those of single mode heating
and mode sweeping heating. The performance of the process control system is evaluated
in terms of processing temperature uniformity, and curing temperature control stability
and accuracy.

Research results are summarized and conclusions are made in Chapter 7.
Recommendations for fiJture research work are presented in Chapter 8. Finally, Hardware
instrumentation and LabVIEW programs are documented in the Appendices. Appendix A
provides the hardware specifications. LabVIEW subprograms used in the characterization
and control programs are documented in Appendix B. LabVIEW programs for system
characterization are documented in Appendix C. LabVIEW programs for process control

systems are documented in Appendix D.

12

CHAPTER 1

MICROWAVE PROCESSING FUNDAMENTALS AND LITERATURE REVIEW

Microwaves are electromagnetic waves with wavelengths measured from 30 cm to
0.3 mm, which correspond to the frequency range 109 — 1012 Hz. The heating effects of
microwaves result from the interactions between material molecules and the
electromagnetic fields. To better understand microwave heating, one needs to know how
electromagnetic fields are established inside the materials, and how they interact with the
material at the molecular level. In this chapter, fiindamental electromagnetic theory related
to microwave processing is presented. The interactions between microwaves and the
materials are discussed. Research efforts in microwave processing are reviewed, including
the recent development in variable frequency microwave processing. Process modeling
and process control issues are also discussed for a better understanding of the microwave
environment for control system development. Throughout this chapter, vectors are

denoted in bold face, and scalar quantities are denoted in italic face.

1.1 Electromagnetic Theory

1.1.1 Electric Field and Magnetic Field

According to Coulomb’s law, the force between two electric charges in free space

separated by a distance r is given by:

12:11:12.7“, (1.1)
47r80r‘

l3

where F is in newtons, r is in meters, ql and q; are in coulombs, u, is the unit vector along

. . . . _ 1
r, and £0 rs the permrttrvrty of free space (=8.854 x 10 ‘2 s: F x10’9 farad/meter). A
it

repulsive force results if the charges are of the same sign; whereas an attractive force
results if the charges have opposite signs.

The electric field intensity E at a point in free space is defined as the Coulomb
force acting on a test unit positive charge placed at that location, assuming a distance r

from the charge q:

E: q .u. (12)
47W“

 

In general, the electric field E at any point in an electrostatic field due to a sum of charges
distributed throughout space can be obtained by summation or integration of the effects
exerted by each charge. The charge distribution can be represented by charge density
p(r) ([=]coulomb/meter3), which is time invariant in electrostatics.

The electric displacement density or electric flux density D ([=] coulomb/meterz) is
defined as:

I) = 80E + P (1.3)
where P is the volume density of polarization ([=]coulombs/meter2), the measure of the
density of electric dipoles.

A static electric charge produces a static electric field, whereas a steady electric
current generates a static magnetic field, which can be detected with a magnetic compass.

A current-carrying wire produces a magnetic field whose direction is related to that of the

14

current by the right-hand rule. This magnetic field creates a force on moving charges in the
field.
The magnetic force acting between two charges moving slowly with constant

velocities is termed the Lorentz force, and in free space can be expressed as:

.V. xu .
E2:flL%wxg;;Ti; (IQ
47: r

where F1; is the force exerted on charge 1 by charge 2, v1 and V2 are velocities of the two
charges respectively, r is the distance between the two charges, um is the unit vector

along r with a direction pointing to charge 1, and ,uo is the permeability of free space

(=4 7r x 10'7 henry/meter).

The magnetic flux density B at a point in free space has the magnitude of the
Lorentz force acting on a test unit moving charge (1 coulomb moving at lm/second) at
that location, assuming a distance r from the charge q moving with a velocity v, but with a

direction perpendicular to both the Lorentz force and the moving charge q:

Therefore, for a charge q. moving with a velocity v. the Lorentz force is:

F=q,V,><B (1.6)
In general, the magnetic flux density at any point due to a sum of moving charges
distributed throughout space can be obtained by summation or integration of the effects
exerted by each moving charge. The distribution of moving charges can be represented by
the electric current density J ([=]ampere/meter2), which is time invariant in

magnetostatics.

15

The magnetic flux density can also be expressed as:

B=pAH+M) 07)

where H is the magnetic field intensity ([=]ampere/meter), and M is the volume density of
magnetization ([=]ampere/meter), the measure of the density of magnetic dipoles in the
material.

A material is called a simple matter if:

Dzdl
szH as)
JzoE

where o (mhos/meter) is the conductivity. 8 = 8,30 , and ,u = ,u,,uo. a, is called the

relative permittivity or dielectric constant, and ,u, is called the relative permeability. In an
isotropic medium, .9 is a scalar having only magnitude. However, in an anisotropic
medium, such as plasma and conductive fiber reinforced composites, 8 is a tensor of rank
two with nine components.

Materials having large a are called conductors and those having small a are
called insulators or dielectrics. Conductors have many “fi'ee” electrons, easily detachable
from atoms. A perfect dielectric material like vacuum has zero conductivity, i.e. no
detachable electrons. All other dielectric materials have finite conductivity. For most linear

matter, the permeability y is approximately that of free space ,uo. There is a class of
materials, called diamagnetic, for which ,u is slightly less than #0 (of the order of 0.01
percent). There is a class of materials, called paramagnetic, for which ,u is slightly greater

than ,uo (again of the order of 0.01 percent). A third class of materials, called

16

ferromagnetic, has values of ,u much larger than ,uo, but these materials are often

nonlinear. Therefore, all materials except ferromagnetic can be called nonmagnetic and

,u = ,1:0 can be taken for them.

A material is called a linear matter if [21]:

DzeE+sla—E—+ez—a—£+
at 612
an 62H
<B:#H+Iul?a7‘+#2—at—2+”' (1.9)
«2
J=oE+ola—E+ozg+
. at at-

 

The physical significance of this extended definition arrives from the consideration of the
mass. The atomic particles of matter have mass as well as charge, so when the field
changes rapidly the particles cannot “follow” the field because of momentum. For
example, there will be a time lag before an electron accelerated by the field can change

direction, when the direction of E changes.

1.1.2 Fundamental Electromagnetic Theory
As discussed in electrostatics and magnetostatics, a charge distribution p(r)

produces an E field, while a current distribution J(r) produces a B field. In a time-varying

case, where there are charge p(r,t) and current J (r,t) in a source region, the Equation of

Continuity holds due to the conservation of electric charge:
6
V~J(r,t)+—é;p(r,t)=O (1.10)
Since p(r,t) and J(r,t) are coupled through the Equation of Continuity, it is reasonable to
think that E(r,t) and E(r,t) are coupled in the time-varying situation.

17

The relations between the field vectors and the source current and charge are

captured by a set of equations called Maxwell’s Equations:

Vsz-aa—It3 (Faraday's law) (1.11)
VxH=J+aa—It) (Maxwell—Ampere law) (1.12)
V-sz (Gauss law) (1.13)
V-BzO (Gauss law-magnetic) (1.14)

where D is defined by Equation 1.3 and H is defined through Equation 1.7:
Hzfi—M or»
,u

Equations 1.11 and 1.12, along with Equation 1.10, constitute the three
independent equations in Maxwell’s theory of electromagnetism. Equation 1.13 can be
derived by taking the divergence of Equation 1.12 and eliminating J between the resultant
equation and Equation 1.10. Equation 1.14 can be derived by taking the divergence of
Equation 1.11 and setting the constant of integration with respect to time equal to zero.
Therefore, Equations 1.13 and 1.14 are dependent equations.

In order to make the Maxwell’s equations definite, we need additional information
provided by the constitutive relations between the field quantities. As aforementioned, in a
simple isotropic medium, the field quantities are related as follows:

Dzdl UIQ

B
H:— urn
,u

18

J = 0E (1.18)

Now the system of Equations 1.10 through 1.12 and Equations 1.16 through 1.18
becomes definite. There are 16 unknowns and 16 equations. In many boundary-value
problems, the constitutive relations between D, B, E, and H are usually known while the
current density J is treated as a source term. In that case, E and H are solved in terms of J
and satisfy certain boundary conditions.

Maxwell’s equations and the Equation of Continuity can also be expressed in
integral form, if the fields and their derivatives are continuous throughout the region of
integration. By integrating the equations through a volume V with an enclosing surface S,

and applying the curl theorem and the divergence theorem, the following equations can be

obtained:
fi(an)dS=-HI%3-dV (1.19)
§(an)dS=m(J+%?—)dV (1.20)
fi(n-J)dS=—m§a§dr/ (1.21)
fi(n.B)dS :0 (1.22)
fim-mdszmpdV (1.23)

In the Maxwell’s equations, equations for E and B are coupled. They can be
decoupled to simplify the solving process of the equations. Through mathematical

manipulations of the Maxwell’s equations, the following equations are obtained [22]:

19

or
at (1.24)

 

 

E}: Vii—Hp

‘3 a 52
v~ _ __._
[ ,uo ye ]{

at t2
8 —,quJ‘

where J’ is the source current term. In a source free region, Equations 1.24 become:

a 62 E
V2 — —— — = 0 1.25
l #0 at #8 612 ]{B} ( )

When the source functions, J(r,t) and p(r,t) , oscillate with a constant angular

frequency of a) in the system, all the fields will oscillate with the same frequency. The
Maxwell’s equations can be written in time-harmonic form:
rV x E(r) = —ia)B(r)

< V x H(r) = J(r) + icoD(r)

V - PM = p(r)
[V - B(r) = o

(1.26)

 

In time-harmonic case, Equations 1.24 become Helmholtz Equations and Equations 1.25

become Helmholtz equations in source-free region:

7 7 o E
[V‘ + anus ]{B} : 0 (1.27)
where
a‘ = 5(1 -i—"—) (1.28)
we

8' is called the complex pemrittivity.

I. 1.3 Boundary Conditions

In the Maxwell’s equations, boundary conditions must be specified at the interface
of two media. For ideal conductors, they can not sustain a field inside. Table 1.1 lists the

boundary conditions associated with the corresponding differential equations [23].

20

Table 1.1 Boundary Conditions

 

 

 

 

Differential Equations Boundary Conditions
VxE:_a_B 1~ “1X(EI“E2):O
at 2_ n,x(E,—E,)=o
3, IIIXEl =0
mm“? .. n.x(H.—H.)=K
t 2- n1X(H1_H2):O
3, DIXH1=K
V.J=2p=o 1. V,.K=—n,.(J,—J,)—°”S
at a:
2. n,.(J,-J,)=-°”‘
61
bps
3. Vs-K=-nl-J,——
a:
V.D:p 1' n1X(Dl—D2):ps
2- “1"(D1—Dzlzps
3- “1XD1=Ps
V-B=0 1. DIX(B,-Bz)=0
2. n,x(B,—B,)=o
3, IIIXBl =0

Case 1. General boundary condition
Case 2. Neither of the two adjacent media being a perfect conductor
Case 3. Medium 2 being a perfect conductor

 

Notes: 1. The unit vector In is pointed from the interface to medium 1.
2. K is the surface current density.

3. p, is the surface charge density.

4. V, is the surface divergence.

21

1.1.4 Electromagnetic Fields in a Cylindrical Cavity

Consider an empty cylindrical cavity (cavity length = h, radius = a) with metal
walls in a time-harmonic situation. The electromagnetic fields can be solved as a function
of position using the Helmhotz Equations 1.27. Since the medium air is virtually vacuum,
8' = e . The equations can be conveniently solved in cylindrical coordinates through
transformation to eigenvalue problems.

The corresponding boundary conditions and assumptions are:

1. Tangential component of the electric field is equal to zero at the cavity

walls, and the top and base of the cavity, i.e. E}, (z = 0, z = h) = 0, 13¢ (r = a, z =

0, z=h)=0, andEz(r=a)=0.

2. Fields must be finite everywhere, which means the Bessel’s function of the
second kind cannot be a solution since it can go to infinite when the argument
goes to zero.

3. Fields must repeat every 27: in ¢.

With the above boundary conditions, the Helmholtz equation can be transformed
into an eigenvalue problem whose eigenfirnctions are Bessel's functions and harmonic
fianctions. These eigenfimctions describe the electric and magnetic field components of a
mode while the eigenvalues index the modes and describe the propagation characteristics
of the mode. There are two different types of modes that can propagate in a cylindrical
cavity, TE (transverse electric) or TM (transverse magnetic). TM modes are solutions to
the Maxwell's equations with the boundary condition that there are no longitudinal

magnetic field components while the TB modes are the solutions with the boundary

22

condition that there are no longitudinal electric field composites. Therefore, in TE modes
the electric field is aligned perpendicular to the direction of wave propagation and in TM

modes the electric field is aligned in the direction of wave propagation.

1.1.4.1 TE Modes

For TE modes the electric field components are transverse and the magnetic field
components are parallel to the direction of wave propagation, which is the z-axis and thus
E2 = O. The TE mode field components in the form of a vector potential is given as [21]:

ann
a

 

w(p,¢, 2) = B...J,.( p)sin(m¢)sin(5,f’-z)

m=0,1,2,3,...
p,n=1,2,3,... (1.29)

The field components are solved as follows [21]:

Ep 2 E11," (2%"i p) sin(m¢) Sim-1% z)

E, = E,J,',,(XT"“ p) cos(m¢) sin(£h7-r-z)

E20

9
6

. X»... pit
Hp : HlJm(-a—p) cos(m¢)cos(Tz)

Hj 2 H21", (Eam- p) sin(m¢) cos(p—: z)

X”!!!
Hz 2 H,J,,( a

 

p) cos(m¢)sin(p—h’r z) (1.30)

where,

23

 

 

E"! X ’E2_J X ’ ' X 7
Ha2
M,... X3 (if?) <13»

1.1.4.2 TM Modes

For TM modes the electric field components are parallel and the magnetic field
components are transverse to the direction of wave propagation which is the z-axis, and
thus Hz=0. The TM mode field components in the form of a vector potential is given as
[21]:

X
Mp. ¢. 2) = A...,,J...( "'"
a

 

p) cos(m ¢) cos(fihj-r— z)

m=0,1,2,3,. . .
p,n=1,2,3,... (1.32)

The field components are solved as follows [21]:

E. = 5,1432%) cos(m¢) amp—Zn
E. = 3,14%,» sin(m¢) sin-3,552)
E. = 5,1,(3‘Zup)cos(m¢)cos(P—:z)
H. = H11... (£51m sin<m¢) cos(p—fz)

. X
H¢ = Hsz(-—a"'i p) cos(m¢)cos(% z)
H :0 (1.33)

2

24

 

 

 

 

(1.34)

1 .1.4.3 Mode Designation

The field component solutions, i.e. Equation 1.30 and Equation 1.33, determine
the characteristic field patterns associated with these modes. These field patterns indicate
the variations and amplitude of the field components as a fimction of the cavity axis. In the

cylindrical cavity the field indices (m, n, p) correspond to the components (o, r, z).
Where r is the radial direction, (15 is the circumferential direction, and z is the vertical

direction. The index m is the periodicity in the radial direction, n is the number of half
wavelengths in the circumferential direction, and p is the number of circular wavelengths
along the vertical axis. For an example, TM) 10 designates a transverse magnetic mode with
(0) wave variations along the radial direction, (1) wave variation in the circumferential

direction, and (0) wave variations in the vertical direction.

1.1.4.4 Electric Field Pattern

The field patterns for different modes can be determined by plotting the magnitude
of the electric or magnetic field components. It is only necessary to plot these patterns for
one plane since similar patterns are repeated in all the repeating planes along the z-axis.
Several plots of the electric field patterns were generated and are shown in Figure 1.1 and
Figure 1.2. These plots are shown as density plots where the white regions represent high

field strength regions and the dark areas represent low field intensity regions.

25

  

imorx)

   

TE(31x)

   

TE(13x) TE(22x)

   

TE(02x) TE(6IX)

Figure 1.1 Electric Field Patterns for TB Modes

26

 

 

TM(01x)

TM(3lx) TM(12x)
TM(22x) TM(03x)

TM(2lx)

 

 

   

TM(32x)

Figure 1.2 Electric Field Patterns for TM Modes

27

v1,

1. 1.4.5 Cut-off Frequency

With the eigenvalues of the solutions to the Helmholtz Equation 1.27, the wave
propagation characteristics can be derived in a form of what is known as the cut-off
frequency equations. The cutoff frequency defines the minimum frequency for a given
cavity diameter that modes can propagate. The cut-off frequency equations for the TE and
TM modes are shown in Equations 1.35 and 1.36. They show a relationship of frequency
as a function of cavity length, It, cavity radius, a, permittivity and permeability, and the
tabulated zeros of the Bessel's fimction or its derivative.

Waves of frequencies below the cut-off frequency of a particular mode cannot
propagate, and power and signal transmission at that mode is possible only for frequencies
higher than the cut-off fiequency. At frequencies below the cut-off frequency of a given
mode the propagation constant is purely imaginary, corresponding to a rapid exponential

decay of the field and the generation of evanescent modes [24].

 

 

 

 

.. ‘ I” l (”I
= "'" + — 1.35
(f)mnp 272‘ #8 J a h ( )
where, x'm =tabulated zeros of the derivative of the Bessel’s fiinction.
(f)”’ ._ 1 (if {ET (1 36)
"mp 2n ya a h i

where, xmn= tabulated zeros of the Bessel’s fiJnction.

28

[ —:r1=b11
l —r1=n12
T1502]

l
l—TErrz
—TE121
——TE122

L—mon

-—TM012

TM021

TM022

, mm

, , TM112

5 10 15 20 25 30 “‘1“:
‘ TMl
L Cavity Length (cm) P77 , ~—~

Frequency (GHz)
4:-

 

 

 

Figure 1.3 Mode Chart for a 7 Inch Diameter Empty Cavity

The significance of the cut-off frequency equations is that they indicate the
frequencies and cavity lengths where a single mode can resonate in a cavity of a known
radius. Another useful form of these equations is a plot of the frequency versus the cavity
length for a fixed cavity radius, to generate what is know as the mode chart (see Figure
1.3). The mode chart shows the locations of modes with respect to other modes as a
function of frequency and cavity radius. It is important to note that these equations can be
used in two forms: by fixing the frequency at a constant value and varying the cavity
length to excite different modes or by fixing the cavity length at a constant value and
varying the fi'equency to excite different modes. The previous method is the frxed

fi‘equency method and the latter is the variable frequency method, which is used in this

29

work. In the variable frequency method more modes can be theoretically excited than in
the fixed frequency method. This is evident by the number of modal lines that a vertical
line intersects for the variable frequency method and the number of modal lines that a
horizontal line intersects for the fixed frequency method.

By intersecting the mode chart, it is apparent that, as the order of modes increase

 

(increase in the indices) the modes become close together in frequency. Thus, suggesting

that it may be difficult to excite higher order single modes. The mode chart also indicates

 

that the modes that can be excited at a fixed frequency.

1.1.4.6 Cavity Quality Factor

The cavity quality factor, or Q-factor, is a measure of how well the cavity stores
microwave energy. The definition of Q-factor of a resonant cavity is defined as the ratio of
the energy stored inside the cavity volume to the energy lost to the cavity surface area per

unit time, as formulated in Equation 1.37 [21].

 

d alel’dr
QZ2¢)£energy store ]:2% —K_'—T— (137)
energy lost RflHl‘ds
S

where R is the intrinsic wave resistance of the metal walls.

The Q-factor is a function of the resonant mode. For an empty cylindrical in
microwave frequency range, it is usually very high with a range fi'om 10,000 to above
40,000 [25]. The theoretical Q value increases as the order of modes increases at a given
frequency. The reason is that for high order modes, the cavity volume becomes relatively
larger, which results in a greater volume-to-surface ratio. Since energy is stored in the

volume and lost on the imperfectly conducting surface, the Q factor increases due to the

30

increased volume-to-surface ratio. In practice, the Q—factor can be lowered by the
introduction of a feed system (coupling system), a large impedance, imperfections in the
construction, and imperfect conductivity at the cavity surface [26].

An experimental method called half power point method is used to measure the Q
factor using an oscilloscope. The power reflectance (input power over reflected power
ratio) curve is generated versus the frequency on the oscilloscope screen. The Q factor is

calculated as follows:

 

A
24f (1.38)

ener stored
=4 . 1

energy lost

The method is illustrated in Figure 1.4.

Power PMAL _ _ __

   

lp__
2

 

frequeIrcy

Figure 1.4 Q Factor Calculation using Half Power Point Method

1.2 Interactions Between Microwaves and Materials

Typical frequencies used in material processing are 915 MHz, 2.45 GHz, 5.8 GHz

and 24.124 GHz. However, if the microwave leakage can be controlled within safety

31

limits, any frequency can be used in industrial applications. When introduced into
microwave field, the materials will interact with the alternating electromagnetic field at the
molecular level. Different materials will have different responses to the microwave
irradiation.

If the material is conductive, the electrons move freely. When exposed to the
electric field, an electric current results. The current in the conductor will heat the material
through resistive heating. However, for conductors with high conductivity, the incident
microwaves will be largely reflected and therefore they can not be effectively heated by
microwave. The fields attenuate towards the interior of the material due to skin effect,
which involves the magnetic properties of the material. The conducting electrons are
limited in the skin area to some extent, which is called the skin depth, d,. Defined as the
distance into the sample, at which the electric field strength is reduced to He, the skin
depth is given by [21]:

d, = 1 , (1.39)

(1 / prop'o)m

 

where a) is the frequency of the EM waves in rad/ sec, no is the permeability of the free

space, 4 it x 10'7 H/m, ,u' is the relative permeability, and 0' is the conductivity of the
conductor in mhos/m. For graphite, or = 7x104 mhos/m, and d, = 38.4 p m at 2.45 GHz

in free space. Therefore, the skin depth for AS4 graphite fiber at 2.45 GHz is about four
times the fiber diameter. As frequency increases, the skin depth decreases.

When a dielectric material is eXposed to electromagnetic waves, four polarization
mechanisms may take place [13]. They are electron (or optical), atomic, dipolar (or

orientational), and interfacial or space-charge polarization. The electron or optical

32

polarization is due to the shift of the orbit center of electron cloud caused by the applied
electric field. No dielectric losses will result since the induced dipole moment is always in
phase with the electric field. The relaxation time for electron polarization is 10'15 seconds
[27]. Atomic polarization results when the molecules placed in the electric field consist of
two different kinds of atoms. The magnitudes of atomic polarization of non-ionic, or non-
partially ionic polymers are much less than those of ionic or partially ionic polymers. The
typical relaxation time of atomic polarization is 10’13 seconds. Orientation or dipole
polarization is observed for dipolar or polar molecules placed in the electric field. The
dipolar molecules will rotate until they are aligned in the direction of the external electric
field. The dielectric loss of orientation polarization is mainly due to the fiiction force that
the dipoles rotate against. The typical relaxation time for this type of polarization is 10'9
seconds. The interfacial or surface-charge polarization is caused by the migration of
charges inside and at the interface of dielectrics in a large scale field. Microwave heating
is mainly contributed by the dipolar polarization, because it occurs in the microwave
frequency range.

The measure of the ability of a dielectric material to absorb and to store electric

potential energy is the complex permitivity,
e = g' — 15' (1.40)
where the real permitivity a" is also called the dielectric constant, which characterizes the

penetration of microwaves into the material. The dielectric loss 8' indicates the ability of
the dielectrics to store the energy. The average microwave power per unit volume

converted into heat is given as follows,

33

PM = 1/2wgog' tan 6E2 (W/mz) (1.41)

where,

(a, + 0, )/ (1)50 + 8'

 

 

tan 5 = (1 .42)
e
is the loss tangent, characterizing the ability of the material to convert the absorbed
microwave energy into heat.
The skin depth for dielectric material is defined as before and given by,
2
d5 = _ , 1],, (1.43)
we (,uo / e)

For fully cured DGEBA/DDS epoxy, e = so (3.5 -j0.1), so the skin depth is d, = 0.729 m

at 2.45 GHz. In general, the skin depth can be calculated for the homogeneous materials

by [281.

 

, 1/2 ‘
1/2 . ~
1 2
d5=_[ , ] [14(2)] —1 (1.44)
2a) popeoe e

The dielectric properties of composite materials are anisotropic. The skin depth

can be calculated in the principal directions using Equation 1.44. For AS4/3501-6
graphite/epoxy composite, the effective complex permitivity is 8’ = 80 (1-j2500) along the
fiber direction, and s’= so (14.5-j75.8) perpendicular to the fiber direction [29]. Taking

the fiber direction as the principal direction, the skin depth of this composite is 9.8 mm

and 3.2 mm for electric fields along and perpendicular to the fiber direction, respectively.

34

The power absorption rate inside the composite can be calculated by Poynting’s theorem

as [21],

I!
H _g.

1 —+
PzaweUEo set, 0E (1.45)

where E,” = Ed + —, 2,1,. and E-d are relative effective dyadic factor and relative
£00)

dyadic loss factor due to dipolar distribution, 5 is the dyadic conductivity in S/m and E'
is the conjugate electric field vector in V/m.

Microwave heating is unique in that it is rapid, instantaneous, volumetric and
selective. Whether a material can be effectively heated by microwave depends on its
dielectric loss factor. High dielectric loss factor results in better heating. Microwaves
offer self-limiting heating in the processing of therrnosets. As the crosslinking proceeds,
the mobility of dipoles decreases due to the “trapping” or reaction. Therefore, the
dielectric loss factor diminishes with the extent of cure. However, in practical
applications, if the dielectric loss factor of the material increases with the temperature,
such as in the rubber processing, thermal runaway will usually be observed. This problem
can be solved by regulating the input microwave power, since microwave heating is
instantaneous. The non-uniformity of microwave heating can also be alleviated by
controlling the electric field exerted onto the material.

Whether there is a “microwave effect” in chemical reactions has been a rather
controversial topic. A proposed mechanism for the “microwave effect” in polymers
suggests a non—equilibrium, non-uniform energy distribution at the molecular level (or a

non-equilibrated temperature), resulting in certain dipoles having a greater energy than the

35

“average” energy of adjacent groups [3 0][3 1][32]. This increased energy corresponds to
an increase in an effective temperature of the reacting groups. The energy couples directly
with a reactive polar group in the system and dissipates through adjacent groups by the
usual mechanisms. It is impossible to determine the efi’ect that microwave processing has
on reaction kinetics based on the available data, because of the range of materials studied,
differences in temperature and control and measurement methods, lack of knowledge on
microwave energy absorbed, and variations in microwave applicators. However, two
general observations can be made [33]: (1) slower-reacting systems tend to show a greater
effect under microwave radiation than faster-reacting systems and (2) the magnitude of the

observed effect decreases as the temperature of the reaction is increased.

1.3 Microwave Processing of Materials

1.3. 1 Microwave Processing of Polymers

Microwaves have been used to process both therrnosets and therrnoplastics. The
heating characteristics for these two types of polymers are different due to the different
dielectric behavior during heating. George et al. used microwave energy to dry various
thermoplastics, including powdered poly(vinyl chloride), pelletized polycarbonate,
pelletized polypropylene and acrylonitrile-butadiene—styrene polymer [11]. The
microwave drying time was about 4-20 times less as compared with the conventional tray
dryer at 104°C. All the microwave-dried polymers have equivalent or higher tensile
strength and impact resistance as compared to the hot-air-dried polymers. Microwaves
have also been used to cure many therrnosets, including polyesters [4][34], polyurethanes

[35][36], polyimides [37][38] and epoxies [4][5][7][9][39][40][41]. Most of the results

36

showed that the cure speed is faster for microwave cure than for thermal cure. Epoxies
are the most widely studied thermosets. Both pulsed and continuous microwave curing of
epoxies have been studied [13][42]. A systematic study was performed to elucidate the
effects of microwaves on the crosslinking of epoxies using two epoxy systems,
stoichiometric mixtures of DGEBA epoxides with a high-loss curing agent, diamino-
diphenyl sulfone (DDS), and a low-loss curing agent, meta-phemylene diamine (mPDA)
[9]. Experimental results showed that increased reaction rates were observed in
microwave cure as compared with those in thermal cure at the same temperature. At low
conversion, the Tg’s were similar for both microwave and thermally cured samples. But
high Tg’s were observed for the microwave cured samples at high conversion. This effect
is more significant with DDS as cure agent than with mPDA, which may be due to the
higher dipole moment of DDS. Bush et al. suggested that the reaction rate enhancement
be the result of a non-equilibrated temperature effect where the local temperature is higher

than that of the bulk, especially for high-loss species such as DDS [3 2].

1.3. 2 Microwave Processing of Composites

In composite material processing, both conducting and non-conducting fiber—
reinforced composites have been processed by microwave energy. Lee et. al. processed
continuous graphite-fiber-epoxy laminates up to 32 plies using a waveguide and
multimode applicators [29]. However, the attempt to process multi-directional graphite-
epoxy composites was not successful. Wei et al. used a cylindrical, tunable single-mode
cavity and successfully processed both cross-ply and unidirectional 24-ply 7.62 X 7.62 cm

Hercules AS4/3501-6 graphite/epoxy prepreg laminates with 2.45 GHz microwaves [18].

37

In comparison, mcirowave processed parts in an unpressurized system after 90 minutes of
heating had similar strength to that of autoclave processed parts with a five hour heating
cycle under a pressure of 690 kPa. Continuous processing has also been studied and
developed for composites with conducting and non-conducting fibers [43 -47]. The
control of microwave leakage at the entrance and exit ports is the critical issue in
continuous microwave processing. Usually an anti-leakage structure called a choke is
used at the entries. However, for conducting fiber-reinforced composites, a specially
designed microwave leakage jacket should be used [48]. In recently developed
microwave-assisted pultrusion processes, the length of the process chamber, the
processing time and the pulling forces were reduced significantly [47] [49][50]. Single-
mode applicators proved to be more efficient than waveguide in the microwave assisted

pultrusion [47].

1.3.3 Kinetics of Microwave Curing of Epoxies and Epoxy Composites

Microwave processing of polymers and composites has been demonstrated to have
advantages over thermal processing, including enhanced polymerization rates [32],
increased glass transition temperatures of cured epoxies [9], improved interfacial bonding
between graphite fiber and matrix [12], and increased mechanical properties of some
composites. Therrnoset epoxy resins are the most widely used matrix materials for
advanced composites. Epoxy resins are processed or cured by conversion of liquid
monomers into a three dimensional therrnoset network via chemical reactions. A large
amount of work has been performed in the microwave curing of the general class of epoxy

resins.

38

 

Thermal cure kinetics models can be used in modeling the reaction kinetics of
microwave cured epoxy resins [13]. There are mainly two types of kinetics models for the
thermal cure of epoxy. One is the nth order reaction kinetics model [56][5 7] and the other
is the autocatalytic reaction model [58-61]. The n‘h order reaction kinetics model assumes
that the kinetics can be expressed as:

do _
E — k(DfM) (1-46)

where a is the extent of cure, t is the time, the function fl a) is expressed as (1- a)“, and

k(T) is the overall reaction rate constant which obeys the Arrhenius relation:

k(T) = Aexp(—§—/T£) (1.47)

From Equations 1.46 and 1.47, one can derive that:

ln(t) — LT}? = [constant] for fixed a or T8 (1.48)

A master curve can be obtained by plotting a or T8 versus ln(t) at a reference
temperature[62]. The activation energy E can be determined by comparing an
experimental curve with the master curve, where the shift should be (E/R)(1/T-1/T,.,f). A
and n can be calculated by fitting the data to the equation:

ln(%?—)+E—;£ = ln(A)+nln(1— a) (1.49)

The autocatalytic kinetics model is more commonly used, in which the
phenomenological cure kinetics expression for a stoichiometric epoxy resin is given
by[58][60][61]:

da .. _ ..
Tit—444.196: )(1 a) (1.50)

39

where k, is the non-catalytic polymerization reaction rate constant, k; is the autocatalytic
polymerization reaction rate constant, m is the autocatalyzed polymerization reaction
order, and n is the non-catalyzed polymerization reaction order. When there is no
etherification reaction or OH impurity, a general cure kinetics expression can be derived as
follows, provided that the reaction rate constants for primary and secondary amine-
epoxies are the same [63]:

62—? =(k1+k2)(1—a)(B— a), when a < age, (151)

where B is the ratio of the initial hardener equivalents to epoxide equivalents. Equation
1.51 works well before the gelation point. However, after the gelation point, the reaction
kinetics is better represented by [64]:

id? = k3(1— a), when a > age, (1.52)

The n‘h order reaction kinetics is computationally simple. According to this model,
the maximum reaction rate should occur at the beginning of the reaction. However, in real
cases, (1 =0.3 ~ 0.4 at maximum reaction rate, which is better explained by the
autocatalyzed reaction mechanism. It was demonstrated that the cure kinetics of
DGEBA/mPDA and DGEBA/DDS systems could be described by the autocatalytic kinetic
model up to vitrification in both microwave cure and thermal cure [13].

In microwave processing, the dielectric property change of materials being
processed becomes an important issue. Comprehensive knowledge of epoxy dielectric
properties during processing is necessary to properly interpret the cure mechanism
induced electromagnetic radiation. Like most therrnosets, the permitivity and dielectric

loss factor of epoxy resins increase with temperature and decrease with extent of

40

cure[65][66]. Epoxy is an efficient absorber of microwave radiation at the beginning of
heating, with 8" increasing as the resin is heated. As the cure reaction progresses, the
heat input will be mainly from the exothermic reaction. The pure compounds of an epoxy
system (epoxy resin and hardeners before mixing) have different dielectric behavior with
respect to temperature. DGEBA (diglycidylether of bisphenol-A) exhibits an a"
maximum around 50 °C [67]. The dielectric constant variations of the hardeners with

temperature is presented in Figure 1.5 [66].

 

 

 

5 0.5 4-
DDS
0.4 ..
0. 3 ~-
DDM
0.2 = ..-‘~~“
/
0.1 4»
0""
y,»
o . a """"""" 4 e
20 40 60 80 100 120
Temperature T (’C)

Figure 1.5 Dissipation Factor 5' vs Temperature for DDS, DDM and DDE

There is a sharp a" increase beyond 150 °C for DDS(4,4'-
diaminodiphenylsulphone) and at about 100 °c for DDM (4,4’-diaminodiphenylmethane)
while no such variation is observed for DDE (4,4'-diaminodipheny1ether). One of the
recent studies suggested that the decrease in the microwave dielectric properties of the

DGEBA/DDS epoxy system undergoing cure is predominantly dictated by the nature of

41

the polymer superstructure, and not as much by the changes in the polar end group
concentrations [68]. It was fiirther concluded that the use of microwave energy for
driving reactions may be usefiil in systems where the network morphology is not rigid and
the dipoles are free to move, giving rise to a relatively high dielectric loss. However, in
another study on dielectric behavior of an epoxy resin during crosslinking at rrricrowave
frequencies, it was found that both the real and imaginary part of the dielectric constant
were mainly affected by the disappearance of specific dipolar species, whose relaxation
times did not change significantly [69].

Microwave processing of epoxy resin composites has been under intensive
investigation among other composites because of their widespread application.
Unidirectional graphite/epoxy composites were first processed up to 32 plies using
microwave energy, while the attempt to process multidirectional samples was not
successful [29]. In later studies, the microwave processing of both crossply and
unidirectional graphite/epoxy was achieved using a cylindrical single-mode cavity [18].
The unpressurized microwave processed composites showed higher modulus with shorter
cure time compared with thermal autoclave process. Part of the reason is that microwave
heating environment can substantially increase the amount of chemical interaction between
the fiber surface and the epoxy resin and amine components of the matrix [12]. As a
result, the composite performance can be improved. Thick section graphite/epoxy
composites were also successfully processed or heated using single-mode cylindrical
cavity [l8][70]. Continuous microwave processing of graphite/epoxy prepregs was also
studied and the processing time was shorter as compared with thermal pultrusion process

[47][71]. Microwaves have also been used to process non-conducting fiber reinforced

42

epoxy composites. A 457 mm long, 127 mm OD epoxy/glass filament wound tube with a
wall thickness of 9.5 mm was processed in one minute using a rectangular multi-mode
cavity at a power level of 20 KW [72]. Different applicators were used to process planar
glass fiber/epoxy laminates [29][73][74]. However, there was no evidence of improved

fiber/matrix bonding by microwaves for glass reinforced composites [12].

1. 3. 4 Other Microwave Heating Applications

Commercial exploitation of microwave heating in the food, rubber, textile and
wood products industries has been successfirl. Modern microwave rubber processing
offers significant advantages over conventional rubber processing, including improved
product uniformity, reduced extrusion-line length, reduced scrap, improved cleanliness,
enhanced process control and automation, and the capability of continuous vulcanization
[2][3][51]. In general, microwave energy enables operating cost reduction, energy saving,
higher quality and reliable products and a greatly improved environment both internally
and externally [2]. Studies have been conducted in using microwave energy to attain the
high temperature required for processing ceramic materials [52]. A combination of
microwave energy with conventional heating has been used to elevate the temperature of
the entire sample rapidly [53]. More uniform microstructures were obtained as examined
as a function of cross-sectional position in the sample. Improved microstructural
uniformity and performance for microwave processed Si3N4 tool bits have been reported
as compared to those processed by conventional methods [54][55]. There has also been
growing interest in application areas such as pollution control, medical sterilization,

medical waste treatment, etc.

43

 

 

1.4 Variable Frequency Microwave Material Processing

Single-mode cavities and multi-mode cavities are commonly used in microwave
heating studies. A familiar example of the multi-mode applicator is the domestic
microwave oven, in which several modes are excited at the same time [14]. Multi-mode
cavities provide somewhat more uniform heating than single-mode cavities if single-mode
cavities use only one mode. However, only single-mode cavities can efficiently couple the
microwave energy into the load, which has been shown in the processing of crossply and
thick-section graphite fiber reinforced composites [75][76]. Single-mode cavities can
provide uniform heating using mode-switching method, in which several modes with
complementary heating patterns are alternatively excited [17]. For a fixed fiequency
system mode switching can be only achieved by mechanically change the volume of the
cavity. This eventually affects the rapidity of the process. The other approach is to vary
the frequency to change modes. The adjustment of frequency is an electronic process. As a
result of the instantaneous switching between modes, not only the speed of the process but
also the controllability of the process can be increased by using variable fiequency
switching.

Variable frequency microwave processing is an innovative method to achieve
uniform heating. The current approach of variable frequency heating is frequency
sweeping. Continuous frequency sweeping method has been demonstrated to be able to
improve the uniformity of heating in multi-mode microwave ovens [77]. By selectively or
continuously sweeping through a certain frequency range within a short time, time-
averaged uniformity of heating can be obtained [78][79]. However, this method has poor
power efficiency and not suitable for processing high lossy and anisotropic materials, in

44

which heating modes usually have common hot spots due to the selective heating of
microwave. By selectively switching through resonant modes, the energy efficiency can be
highly improved. To ensure uniform heating pattern, the heating cycle can be programmed
such that the heating times of different modes are weighted by the corresponding heating

characteristics.

1.5 Process Modeling and Processing Control in Microwave Processing

1. 5. 1 Microwave Process Modeling

The study of microwave processing of materials is interdisciplinary in that it calls
upon the expertise in electrical engineering, material science, process engineering and
design. To enhance its potential industrial applications, a large amount of effort has been
expended in the studies to better understand and design the process. Successful models
always help the design and control of real processes. Microwave process modeling
involves the prediction of the electromagnetic field and the temperature distribution inside
the material with the occurrence of chemical reactions and rheological changes. Usually,
the extent of cure as a fiinction of time and space is desired so as to predict the properties
of the final product.

In general, a microwave material processing model should include mass balance,
energy balance, and kinetic equations. Mass balance takes resin flow in the composite into
consideration. Energy balance involves heat transfer, reaction heat, and microwave energy
absorption. The kinetic equations are used to compute reaction rates and predict the
curing time. These equations are coupled and therefore present a very complex system of

equations. An example of the coupled system of equations are given as follows:

45

(1) Mass Balance:

dM
.:_ Av 1.53
dt p. c ( )

 

 

where dtc rs the rate of change of mass M c m the composrte, pa rs the resrn

density, and A is the cross-sectional area. v6 is the velocity of the resin flow in the

 

composite and can be represented by Darcy's Law[80]:

v = 111.8 (154)
,u dx

where S is the apparent permeability, ,u is the viscosity, and g: is the pressure
gradient.
(2) Energy Balance

0 6 6T '

— T =—K——— + H+P 1.55

at(.0C,,)ax(6x)/) .. ()

where p , Cp, and K are the density, the specific heat, and the thermal conductivity

of the composite respectively. H is the rate of heat generation per unit weight by
the chemical reaction, which needs to be calculated with the aid of the kinetic
equations. P... is the rate of heat generation per unit volume by the absorbed
microwave energy. To compute Pm, the Maxwell's Equations need to be solved for
the field distribution inside the composite material.

(3) Kinetic Equations

The kinetic equations for Hercules 3501-6 resin are given as an example [81]:

46

{—19,- = {(k1 +k2a)(1—a)(B—a) a < age, (1.55)

dt k3(1 _ a) a > age!

where a is the extent of cure, k1, kg, and k3 are the reaction rate constants, B is the

ratio of initial hardener equivalent to epoxide equivalents, and a is the extent of cure at

gel
the gelation point.

Various models for computing the electric field inside a material placed in a
microwave cavity have been proposed. To calculate the electric fields inside a microwave
cavity, Maxwell’s equations need to be solved with appropriate boundary conditions.
Either an integral or differential formulation of Maxwell’s equations can be used to start
with. Analytical approaches have been used in simple cases of loaded rectangular
applicators, such as the equivalent circuit [82], the moment method [83], and the matching
method [84]. A cavity perturbation technique is usually applied when the cavity is loaded
with a small object, which only perturbs the resonant frequency by a few percent [21].
This approach has been used in the measurement of dielectric properties of polymers in
TM012 mode [42]. For a cylindrical cavity coaxially loaded with a homogeneous, isotropic
lossy rod, the electric field inside the cavity was calculated using the mode-matching
technique [85][86].

In general cases where the loaded samples have arbitrary geometry and anisotropic
dielectric properties, one can only resort to numerical methods. A three dimensional finite
element method (FEM) was used to calculate the power dissipated in lossy materials in a
short-circuited rectangular waveguide [87]. Also develOped was a three dimensional FEM
to simulate the microwave field structure and the associated power distribution in the
dielectric material in a multi-mode rectangular cavity excited by waveguides [88]. The

47

numerical results were in good agreement with the analytical solutions. This method is
applicable to microwave heating with arbitrary geometry of the cavity, load and varying
dielectric properties of the load. However, the three dimensional finite element methods
solving for electric fields require large computer storage and long computing time,
especially when small FEM mesh sizes are needed to ensure desired accuracy. For
example, the higher the permittivity of the material, the smaller the mesh size because of
the decrease of the wavelength inside the material.

Composite materials usually are highly anisotropic. Various studies have been
conducted on the interaction between traveling TEM waves and graphite fiber/epoxy
composites [29][89][90][91]. In order to approximate the microwave absorption rate
during the processing of anisotropic composite plates inside a cylindrical cavity, a
simplified five-parameter model was developed by Wei et al. [13][92]. The incident wave
on composite surface is assumed to be a linearly polarized transverse electromagnetic
(TEM) wave. The power dissipated in the composite is decomposed into two
components, P1 and P2. P1 is the dissipated power due to the incident TEM waves from
the sides. And P2 is the power due to the incident TEM waves on the top and the bottom.
P1 is finther assumed to be constant through the thickness of the composite. For a
composite lamina, E in each laminate is composed of the electric fields propagating in
opposite directions:

E=l/2(E’,.+E’,..1)+1/2(E',,+E'n.1) (1.56)
where the TEM wave in each direction is obtained by averaging the waves at the two

interfaces.

48

For TEM waves in the composite material, the electric field can be decomposed
into its two principal directions, parallel and perpendicular to the fiber. The attenuation of
the electric field in each direction can be computed separately. Therefore, the TEM
waves propagating in two opposite directions can both be expressed in terms of the waves
in the two neighbor laminates.

A system of equations can be obtained by combining the equation for each
laminate. The system can be solved if the incident TEM waves form the top and the
bottom are known, as well as the power dissipated due to these two incident waves.
Therefore, the total power dissipation can be calculated by summing the power dissipated
in each laminate due to the top and bottom incident waves and the power dissipated due to
the side incident waves. A five parameter expression for the total power dissipation can
be obtained and the five parameters can be optimized by minimizing the error function for
temperature measurements [13]. A FORTRAN code combining the energy balance
equation, the microwave power absorption model, and the least squares optimization was
developed to generate the five parameters based on the temperature/time/position profiles
obtained during microwave heating of a fully cured composite. With the relationship of
the five parameters as functions of input power for the given mode, the temperature

profiles during the microwave cure can be readily simulated [13].

1. 5.2 Microwave Process Control

Thermal cure profile during material processing is the major factor in consistent
and high quality manufacturing [93]. The main goal of the microwave process control

system is to achieve an optimal thermal cure profile. Different composite materials require

49

different temperature cycles. However, there are two invariant requirements for all optimal
thermal cure cycles. They are uniformity of processing temperature and the accuracy of
temperature control so that the desired thermal cycle can be realized. Therefore, these are

the main objectives of a microwave process control system.

 

Since microwave processing of polymer composites is still in its developing stage,
the concept of control is novel. The first published automated single-mode resonant
cavity was that developed by Alliouat et al. [94] for sintering ceramic materials. The

control system was based upon elements of intelligent control for regulating the input

power and for tuning the cavity. A gradient search method was used for tuning the cavity
where only sensed information about the cavity length and reflected power were required.
Components of this processing include an infrared pyrometer for measuring the surface
temperature, detectors for sensing the input, reflected and absorbed power. Controlled
parameters were the microwave power supply by analog output, and stepper motors for
adjusting the cavity volume.

low et al. [7] developed a controlled pulsed microwave processing and diagnostic
system to cure expoy/amine resins at 2.45 GHz. Temperature excursion resulting from the
exothermic reaction was effectively eliminated and the epoxy/amine resins were cured at a
higher cure temperature without degradation. This system was also capable of measuring
dielectric loss factor during the controlled pulsed power cycle. However, the accuracy and
precision of the temperature control needed to be improved.

Adegbite et al. [20] developed an automated single-mode cavity in order to
advance it as a viable process. In the automation, a control system was designed and built

in addition to the development and implementation of a set of sophisticated and

50

comprehensive control software programs for controlling the curing process in the cavity.
These control programs combine traditional and non-traditional control methodologies.
The control software programs were developed for mode tuning, mode selection and
power control which were constructed independently and then integrated to form the
overall closed-loop feedback control system. A mathematical 2—dimensiona1 simplex
method was used to construct the tuning control software. Both coupling probe depth
and cavity length were adjusted simultaneously to tune the cavity. A traditional PID
(proportional-integral-derivative) methodology was used for the power controller. The
diagnostics system was also automated to provide for automatic empty cavity
characterization and for automatic dielectric analysis of materials inside the cavity.

Beale et al. [95] designed and simulated a temperature control system for the
process of microwave joining of ceramics in a single mode cavity. A microwave heating
model was used in the control system design with the dielectric located at the maximum
electric field position of a rectangular T E103 mode. The heating model equation was
linearized and a closed control loop with a compensator was designed. The control
algorithm was designed with the complete nonlinear model representation of temperatures
at the surface and material center and the temperature dependence of the loss factor and
cavity quality factor. The computer simulation yielded good results for the desired values
in the material temperature, giving a zero steady-state error, closed-loop stability, and on
overshoot of the desired temperature. The simulation results showed that the control
system was able to overcome the thermal runaway problem associated with microwave
joining of ceramics. Since no experimental results were presented in the paper, the

applicability of the control system remains to be proved. The modeling of microwave

51

joining given in this paper appears to be less complicated than the modeling of polymer
composite processing in a microwave cavity.

Research efforts have also been made in developing a knowledge-based system for
the control of microwave composite curing [20][96][97][98]. The problem-solving
architecture addresses the entire life cycle of composite-materials fabrication from a
generic-task viewpoint. The prototype systems capture the experience-based static design
of fabrication plans, and the process-control knowledge of cavity tuning for the
microwave curing of composites. The knowledge-based system for designing microwave
composite fabrication plans will be beneficial when the application of microwave
processing in industry becomes common. In order to achieve wide application of
microwave processing technology in composite industry, process control systems at the
material processing level need to be first developed with consistent and good performance.

The implementation of control strategy is limited by the sensing technology in
microwave processing. For material processing in a single-mode cavity, the input and
reflected power are measured using the power meters, which are used to tune the cavity
and regulate the input power respectively. The temperatures are usually measured by the
fiber optic therrnometry. The desired process parameters such as the electric field strength
inside the material and the dielectric properties of the material can not be measured on-

line.

52

CHAPTER 2
VARIABLE FREQUENCY MICROWAVE PROCESSING SYSTEM AND

COMPUTER CONTROL INSTRUMENTATION

In this Chapter, the components and configuration of the variable frequency
microwave processing system are presented. The hardware instrumentation of the
computer control system is also discussed. The variable frequency system has a power
source with adjustable frequency, thus requires microwave components that operate in the
corresponding frequency range. Microwave frequency and power are the two control

parameters used to achieve optimized and uniform processing.

2.1 Variable Frequency Microwave Processing (VFMP) System

The configuration of the VFMP system is shown in Figure 2.1. It has the same
configuration as the fixed frequency microwave processing system. However, the
microwave circuit components now have an operating frequency range instead of a single

operating frequency.

2. 1. 1 Variable Frequency Microwave Power Source

The variable frequency microwave power source consists of an HP83 SOB Sweep
Oscillator, an HPTM 8623 5A RF plug—in and a LambdaTM VariWave‘D microwave power
source. The LambdaTM VariWave" microwave power source was used as a TWT
(Traveling Wave Tube) amplifier. The oscillator and RF plug-in firnction as the low power
signal generator. The frequency can be manually or automatically adjusted from 1.7 GHz

to 4.3 GHz. The power output from the RF plug-in was adjustable from 6 to 16 dBm. The

53

amplification ratio of the TWT amplifier was high enough to generate a microwave output
of 150 watts. Both the power output of the RF plug-in and the amplification ratio of TWT

change with frequency.

2.1.2 Cylindrical Single-Mode Resonant Cavity

The cylindrical single-mode resonant cavity was made of brass. A schematic of the
cavity is presented in Figure 2.2. The cavity has a diameter of 7 inches with cavity length
adjustable from 10 cm to 30 cm. The coupling probe is side mounted 1.2 inches above the
base of the cavity, with probe depth adjustable fiom 0 cm to 50 cm. The cavity length was
adjusted by moving the top plate. The bottom plate of the cavity was removable so that
sample could be loaded. Both the top and the bottom plates were shorted with the cavity

wall by metallic finger stocks.

    

Coupling Probe

 

 

 

 

     
 

 

 

 

 

 

Pacillatorl Plug-in Pi. Amplifier
f 1

  

directional
couplers

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l
: :
1 wer ‘_ l
: i‘" i W :
l
E l I l 'n l l
O : : ! SCI OSCOPC '7," l :
: 1—1
I ' 1 |
1 PC ' '
L _ ‘[ 11:12 .. I _ __V —————————————————— j V
RS-23 2C
Fluoroptic
Thermometer

Figure 2.1 Variable Frequency Microwave Processing System

54

 

 

 

 

 

 

 

 

Coupling Probe Shorting Plite ,

 

 

 

 

 

' Cl Lc

l

Figure 2.2 Cylindrical Single-Mode Resonant Cavity

 

 

 

 

 

2.1.3 Other Microwave Circuit Components

Other microwave circuit components included the circulator, directional couplers,
power meters, the oscilloscope, and the dummy load. The circulator was used to protect
the microwave power source from the microwave power reflected back from the
microwave applicator. It redirected the reflected power to the dummy load. Power meters
were used for both input and reflected power measurement. Directional couplers were
used to obtain microwave signal in the measuring range of the power meters. Oscilloscope
was used for low power diagnostics of the resonant cavity, which will be discussed in

Chapter 3.

2.2 Automation of the Variable Frequency Microwave Power Source

The automation of the variable frequency microwave power source was essential
to the development of the process control system, because microwave frequency and
power were the only controlled parameters. Instrumentation details are presented in the

next section.

55

2.3 Computer Data Acquisition and Control Implementation

The goal of the process control system is to achieve optimal product quality, which
is greatly dependent on the temperature history of the materials being processed.
Therefore, temperature control is essential in polymer composite processing. If
temperature is too low, the polymerization reactions may not happen or proceeds very
slowly. On the other hand, if the temperature is too high, undesired chemical reactions will
occur and result in defective products. In addition, uneven temperature distribution inside
the material leads to residual stress and affects the strength and other properties of the
product. The temperature distribution can be controlled by changing the microwave
fiequency, while the temperature level can be varied by adjusting the microwave power.
There are different approaches to achieving uniform processing temperature at a desired

level, as will be discussed in details in the chapters to follow.

2. 3. 1 Measurement Instrumentation

2.3. 1. 1 Temperature Measurement

Since the samples were placed in the microwave field when being processed, the
measurement of temperatures requires sensors that were transparent to microwave. Two
temperature measurements systems that used optical fibers were used in the studies. One
is the LuxtronTM Fluoroptic Thermometer, the other is the NortechTM F iberoptic
Temperature Measurements Unit. The sensors of both systems are probes consisted of
optic fibers at the core coated with low dielectric loss polymeric material. At the tip of the
probes was a phosphor sensor with a fluorescent decay time that changes with

temperature. The temperature change was detected by monitoring the percentage of

56

infrared light reflected back through the optic fibers. The probe tips of the thermometry
were inserted through the mold and placed on the top of the sample. Therefore, only

surface temperatures were measured.

2.3. 1.2 Power Measurement

Input and reflected microwave powers were measured using directional couplers
and microwave power meters. Directional couplers are used to scale down the microwave
power signal to within the power meter measurement range. Through the power meter,
microwave power level was converted to analog voltage signal. The voltage signal value

was converted back to microwave power level by the computer program.

2.3.2 Control Instrumentation

2.3.2.1 Frequency Control

The variable frequency microwave power source was consisted of an HP
Oscillator with RF Plug-in, and the LambdaTM VariWaveO microwave furnace as an
amplifier. Since the I-IPTM Oscillator was the signal generator, frequency control was
achieved by controlling the Oscillator output frequency. The GPIB interface of the HP
8350B Sweep Oscillator provided the communication interface with the computer. Any
frequency change can be carried out digitally through GPIB. Either single frequency mode
or frequency sweeping mode could be selected. The instruction string to set the frequency
started with "CW" followed by the frequency. For example, if the desired frequency was

2.45 GHz, the computer would write "CW2.45" to the GPIB port.

57

2.3.2.2 Power Control

The control of the microwave power was a little more complicated issue. There
was only a manual dial knob that could be used to adjust microwave power. One was
installed on the RF plug-in, and the other was installed on the LambdaTM VariWave’1D
power source. Two different approaches were used to achieve computer control of the
microwave power. The first one was using a stepper motor to adjust the manual knob on
the LambdaTM VariWave® to change the amplification. The pin configuration of the cable
connector for the stepper motor is presented in Appendix A.

The other approach was using a variable attenuator with voltage-controlled
attenuation rate. The variable attenuator would be connected between the signal generator
and the amplifier since it has a survival input power of only 30 dBm (1 watt). The
connection configuration of the variable attenuator is also presented in Appendix A. The

characteristics of these two control units are examined and discussed in Chapter 2.

2. 3. 3 Computer Data Acquisition

The configuration of the connector board is given in Appendix 1 along with the
configuration of the National InstrumentsTM PCI-MIO-16XE-50 data acquisition board.
Inputs were the analog signals from the LuxtronTM and the NortechTM temperature
measurement units, and the power meters. The outputs are the control voltages to the
variable attenuator and the stepper motor. The A/D and D/A conversions were
accomplished through the National InstrumentsTM PCI-MIO-16XE-50 data acquisition

board.

58

CHAPTER 3
CHARACTERIZATION OF VARIABLE FREQUENCY MICROWAVE

PROCESSING SYSTEM

In this chapter, characterization results for the variable fiequency
microwave processing system are presented. Mode curves were measured for the empty
microwave cavity and compared with theoretical calculations. The technique of
characterizing the microwave cavity loaded with composite samples was also discussed.
Mode fi'equencies, at which there was effective heating, were located. The heating
patterns of the modes were obtained. This characterization technique is essential to all the
processing experiments. The response time of the power meters was tested and the

characteristics of the variable attenuator was also determined.

3.1 Variable Frequency Method

The theoretical characteristics of the electromagnetic field in a cavity can be
determined by solving the Maxwell’s equations with appropriate boundary conditions
[21]. The cut-off frequencies corresponding to resonant modes can also be calculated for a
fixed cavity volume. Theoretically there are two factors that determine a resonant mode in
a cylindrical cavity: the frequency and the cavity length. Consequently, two approaches
can be used to select a particular mode, as seen in Figure 3.1. Variable frequency method

changes the mode electronically, thus is much faster than the fixed fiequency method.

59

fixed fi'equency approach

variable frequency approach

Mummy (GHZ)

 

5 10 15 20 25 34-—m121
, . TM122
Cavrty Length (cm) *_ .. J

 

 

Figure 3.1 Mode Chart of the Empty Cavity

As fiequency changes, not only the field pattern changes, the coupling efficiency
of microwave energy into the load changes too. For a homogeneous, non-magnetic

material, the microwave power absorption rate, P, (in W/m3) can be modeled as following
[21]:

(3.1)

 

 

where E is the electric field strength inside the material, V/m,
a) is the radian frequency, rad/sec, w=2 I: f, f is the frequency in Hz,
80 is the free space permittivity, so =1/(36 7r)x10'9 F/m,
e" is the effective relative loss factor, a" = and + a/( £0 a) ),

60

I.
e d is the relative loss factor due to dipolar contributions,

and a is the material conductivity.

ll
Qualitatively, assuming a .1 and 0' are constant, as frequency increases the power

adsorption rate P increases for the same E . Therefore, if the cavity is firlly loaded with
homogeneous non-magnetic material the coupling efficiency increases with fi'equency.
However, this model is not applicable when the material is anisotropic, which is
almost always the case. From the mode chart in Figure 3.1, we can see that as frequency
increases the order of resonant modes are getting higher. Since the E-fields of higher order
modes are not as concentrated as lower order ones, the coupling of microwave energy is
getting less emcient as the frequency increases when the cavity is partly loaded.
Computation intensive model should be used to relate the fiequency to microwave energy
coupling. As for a fiber-reinforced composite, a five-pararneter microwave power

adsorption model is available in literature [13].

3.2 Variable Frequency Microwave Power Source Characteristics

As aforementioned, the variable frequency microwave power source consisted of a
signal generator and a TWT amplifier. The frequency of the microwave output could only
be controlled through the signal generator. However, the microwave power level could be
controlled either through the signal generator by adjusting the attenuation level, or
through the TWT amplifier by tuning the dial knob. In computer control implementation,
the power could be controlled through variable attenuator connected to the signal

generator, or the stepper motor attached to the dial knob of the TWT amplifier.

61

It was observed that the microwave power level changed when the frequency
changed, even though the position of the dial knob for power level adjustment remained
unchanged. As discovered by experimental tests, when the fi'equency changed, the power
level of signal generator output changed. Moreover, the amplification rate of the TWT
amplifier changed with frequency. Therefore, if the microwave power were to be
maintained at constant when frequency was changed, a power control action would be
needed. The microwave power level versus frequency curve is presented in Figure 3.2.
The power level was about 25 watts at 2.45 GHz. The curve would change if power level
changed. Generally, microwave power increased as fi'equency increased. Microwave

fi'equency and power also shifts slightly with time.

 

 

90

E80-

ON
C
1

 

 

Microwave Power Source Output

 

 

—‘ N W -5 U!
0 O O O O O
1 1 1 1 1

 

I I I I I I I I I

2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4
Frequency (GHz)

N

 

 

 

Figure 3.2 Microwave Power Source Output Versus Frequency Curve

62

The frequency of the variable frequency power source was controlled through the
GPIB interface between the HP Oscillator and the computer. The speed of frequency
control, i.e. the maximum frequency change rate, was very important to the variable
frequency microwave processing studies. For example during the variable frequency
microwave processing experiments, mode tuning, in which the power reflectance was
measured while the frequency was swept in a certain range, needed to be carried out as
fast as possible. Given sweeping time and interval, the size of the fiequency sweeping
range depends on the speed of frequency control. In order to measure the speed of
frequency control, a program was written to write a number of fi'equencies to the
Oscillator continually, and the total time used was measured. The time used to write a
single frequency was obtained by dividing the total time by the number of frequencies
written. Table 3.1 shows the test data. The result was that it took approximately 0.1

seconds for each fi'equency write.

Table 3.1 Time for the Computer to Write Frequency to the Oscillator

 

Tests Sweeping Range Sweeping Number of Total Time Time for Each
Stg) (GHz) Writes Used (seg Write (sec)

 

 

 

 

l 2 GHz to 4 GHz 0.001 2001 196.80 0.09835
2 2 GHz to 4 GHz 0.01 201 19.81 0.09856
3 2GHzto4GHz 0.1 21 2.13 0.10143

 

 

 

 

 

 

 

63

3.3 Characterization of the Empty Cavity

To test the theoretical predictability of the cavity, mode measurements were done using
swept frequency diagnostic method. Mode fiequencies were determined by identifying the
valley of the power reflectance versus frequency curve. Mode curves were obtained
experimentally by changing the cavity length while recording the mode frequency. Linear
regression was carried out using the linear relationship between p2 and x2, obtained by
manipulation of the cut-off frequency equations (Equation 1.35 and Equation 1.36).
Therefore p and x were obtained and the mode was identified. A comparison of theoretical
and measured mode curves for TM012 mode is presented in Figure 3.3. Measurement
results are shown in Table 3.2. The experiment data were consistent with the theoretical
results. The electromagnetic behavior of the cavity is theoretically predictable. Therefore

the cavity satisfied the requirements for a single mode resonant cavity.

Table 3.2 Experimental Measurement of Resonant Modes in an Empty Cavity

 

 

 

Measured TM012 TM211 TMlll TM112 TE211 TE311
“‘Odes /TE011 fTEOlZ

% error ofp 2.3% 0.2% 4.3% 1.7% 3.4% 6.1%

% error of x’ 3.0% 0.4% 0.4% 0.3% 0.0% 0.1%

 

 

 

 

 

 

 

 

 

* xis the corresponding root of Bessel’s function (TM modes), or the derivative of

Bessel’s fimction (TE modes).

64

 

 

 

+exp. data

+ theoretical curve

 

 

 

cut-off frequency (GHz)
N
in

 

 

 

1 1 t t 1 t 4 t :
8 10 12 l4 l6 18 20 22 24 26
cavity length (cm)

 

 

 

Figure 3.3 Comparison of Theoretical and Experimental TM012 Mode Curves

3.4 Characterization of the Loaded Microwave Cavity

The characterization of the loaded microwave cavity included locating the
empirical mode frequencies and obtaining the heating pattern of each empirical mode.
Empirical modes were determined by locating the local minimums of the power reflectance
versus frequency curve for the loaded cavity with fixed cavity length and probe depth. A
LabVIEW program was developed to acquire the power reflectance versus frequency

curve and the details of the program (characterization&temp.vi) can be found in Appendix

C.

65

 

 

0.9 —
0.8
0.7 —
0.6
0.5

0.4 n
0.3 - N u
0.2

0.1

l 1
£-

1

Preflected/Plnput

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

T I I I I I I l

2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4

 

 

Frequency (GHz)

 

Figure 3.4 Power Reflectance versus Frequency During Frequency Scan

The characterization procedure for loaded cavity consisted of the measurement of
input microwave power, reflected microwave power, and temperatures while sweeping the
frequency from 2 GHz to 4 GHz. This frequency range was used instead of the available
range fiom 1.7 GHz to 4.3 GHz, because the maximum microwave power output outside

of the range fiom 2 GHz to 4 GHz was not high enough for composite processing.

66

 

III
C

 

A
M

A
O

 

 

 

Te mpe rature (°C)

 

 

 

2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4
Fre que ney (G Hz)

 

 

Figure 3.5 Temperature Change versus Frequency During Frequency Scan

Examples of the characterization results are given in Figure 3.4 and Figure 3.5.
Due to the input microwave power, the temperature of the sample would rise during the
frequency sweep. Figure 3.5 shows the temperature profiles versus frequency. As seen
from the figure, the beefing rates were changing as frequency changed. The reason was
that at different fiequencies, the electromagnetic fields inside the sample were different.
As a result, the heating preference inside the sample differed. From Figure 3.5, important
information can be obtained - how the heating preference changed at different frequencies.
The sample could be loaded at different positions inside the cavity and the heating
preference change with frequency would be different. An optimal loading position could

be determined by measuring the temperature versus frequency profile and choosing the

67

position that would have the most diverse heating preference. This means that if different
fi'equencies were excited for heating, the sample would have the most uniform time-
averaged temperature distribution when loaded at the optimal position.

Figure 3.4 presents the power reflectance versus frequency curve. The frequencies
at which the power reflectance is locally minimal and close to zero were considered the
frequencies of empirical modes. Like the theoretical modes, these empirical modes had
very small reflected power versus input power ratio. Some of these empirical modes
correspond to theoretical modes, although the frequencies are different. Some others may
be the result of two or more modes merging together due to the disturbance of the sample
to the electromagnetic field. The power reflectance curve is not exactly reproducible due
to the slight variation of microwave fiequency and power versus time. However, the
variation is usually less than 1%.

No attempts were made to correlate empirical modes to theoretical modes because
the fields have been altered dramatically due to the lossy sample. An empirical mode
would usually have a drastically different E-field distribution that its corresponding
theoretical mode, if there is a corresponding theoretical mode. Therefore, any information
regarding mode heating characteristics needed to be determined experimentally for it to be
utilized during the processing. From here on, empirical modes will be just mentioned as
modes. Therefore, the modes in a processing context are different from the modes
mentioned in a theoretical context.

After the frequencies of the modes were determined, samples would be heated at
these frequencies and temperatures were measured. The heating patterns of the modes

were thus obtained. The control system then will use this information during the

68

processing to select which mode to heat. There are different approaches to applying the
heating characteristics in the mode selection process. They will be further discussed in the

next few chapters.

3.5 Variable Attenuator Characteristics

Variable attenuator is an attenuator with adjustable attenuation by voltage control.
It was used in the microwave circuitry in order to provide a means for the computer to
control the microwave power level. The computer sent out an appropriate control voltage
to attenuate the power of microwave signal from HP oscillator to a certain level before
being amplified by the Traveling Wave Tube applifier. The specifications of the variable
attenuator provided by the manufacturer is given in Appendix A.

The specifications of the Variable Attenuator indicate that the attenuation of the
attenuator is proportional to the control voltage. The attenuation range is 0 to 60 dB, and
the corresponding control voltage irange is 0 to 6 volts. In order to verify this relationship,
attenuation measurements were made while changing the control voltage. The LabVIEW
program (vatest.vi) for the characterization of the variable attenuator is documented in
Appendix C. The test procedure was as follows:

1. Set control voltage = 0;

2. Measure the initial microwave power output from the power source (after the

amplifier) = P0;

3. Set control voltage = v;

4. Measure microwave power output = P;

69

According to the specifications (see Appendix A), the plots of 10Log(P/Po) versus
v should be straight lines with a slope of — 1. The results at five different frequencies (2
GHz, 2.5 GHz, 3 GHz, 3.5 GHz, and 4 GHz) are presented in Figure 3.6. Linear
regression was used to fit the curves to linear equations. Overall, the linear fit was a close
representation of the attenuation vs. control voltage relationship. For example, at f = 3
GHz, the slope of the linear fit was —1.0082 with an R2 value of 0.9937. However, when
the control voltage gets close to zero, the curve is not linear anymore. The linear
relationship between the attenuation and the control voltage holds when the control

voltage is greater than 0.2 volts. When the control voltage is smaller than 0.2 volts, the

 

 

 

 

 

 

 

 

attenuation is very close to 0 dB.
0 , 4.
1
Linear Fit for f= 3.0 GHz
-5 . y = -1.0082x + 2.497
1?.2 = 0.9937
A -10 = \
é . \
e \ \
3 x
'1 -15 = —-+-—f=2.0GHz ‘~{\\
--o---f= 2.5 GHz ‘~:\‘
-—-—- r= 3.0 GHz ‘\Z‘ 4._
'20 ° ----—r=3.5 GHz \,
=--—.----— f= 4.0 GHz
-25 I I I I T I T I I _I I I T I I I I I I
0 0.5 1 1 5 2
Control Voltage (V)

 

 

 

Figure 3.6 Relationship between Control Voltage and Power Attenuation

70

3.6 Power Meter Response Time

In the processing experiments of this research, step power changes were very
common. They were mostly due to one two actions, frequency change or attenuation
change. Because the microwave power output from the power source varied with
frequency, if there was a frequency change, a microwave power change would follow. In
microwave power control, the attenuation of the variable attenuator was changed to adjust
the microwave power. The discontinuous change of the attenuation also would result in
step power changes.

Since the power meters featured a dial attached to a spring, the dial movement
could not precisely follow sudden power changes because of inertia. Therefore, the power
reading immediately after the power change was inaccurate. A finite amount of time was
needed for the mechanical dial to settle after power changes. In order to obtain accurate
power measurement, the response time of the power meters needed to be determined.

Experiments were carried out to test the power meter response by changing the
attenuation of the variable attenuator. The frequency was fixed at 2.5 GHz. Power
changes of different step size were used to characterize the power meter response. The
LabVIEW program (p-response-test.vi) is documented in Appendix C. The test was
carried out as such:

’ 1. A step change of microwave power was made by changing the control voltage
of the variable attenuator.

2. The computer measured the microwave power by acquiring the analog output

from the power meter at discrete times with an interval of 20 ms.

71

 

 

Measured Power (W)

 

10 30

 

50 70 90 110 130 150170190 210 230 250 270 290

Time (ms)

Power Step
Change (W):
[__._—E]
+ 20
+ 30
—-x— 40
+ 50
—Q— 60
—1— 70
_._ 80
_._ 90

 

 

 

Figure 3.7 Measured Microwave Power versus Time after Power Step Change

 

 

Power Difference Percentage

l
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1

0

 

 

10 30 50 70 90 110 130 150 170 190 210 230 250 270 290

Time (ms)

Power Step
Change (W):
-+— 10
—O— 20
30
+ 40

+50,

 

+60
—+—70
—-—-80
*90

 

 

 

 

Figure 3.8 Difference Percentage versus Time after Power Step Change

72

' - I
0 n
-
I -
l
.
7
.
D
. .
. /
. .
. r
J
v
I
a
C
‘3
9.
e
4;:
.
.,
‘ l: 1- I
'8.
. a]:
,.

The results are presented in Figure 3.7 and Figure 3.8. As we can see in Figure
3.7, the analog output of the power meter did not reflect the steady power level until some
time after the control voltage change. Therefore, during microwave power adjustment, the
computer should wait for a certain amount of time after attenuation change in order to
assure accurate power readings. The percentage of power difference between measured
power and steady power was plotted versus time in Figure 3.7. About 170 ms after the
attenuation change, the measured microwave power was within 10% of the steady power
even for power step change of 90 watts. The power difference percentages at different
power step changes after 170 ms are listed in Table 3.3. The power difference percentage

was calculated as follows:

Riteadv — P measured
Power Difference Percentage = = ‘ x 100% (3.2)
stead)»

 

Usually, the power change during processing would be less than 30 watts. In those cases,
the measured power would be within 5% of the steady power 170 ms after the attenuation

change.

Table 3.3 Power Difference Percentage after 170ms.

 

Power Change 10 20 30 40 50 60 7O 80 90
Step (W)

PowerDifference 0.3% 1.3% 3.5% 3.5% 4.9% 5.5% 8.4% 8.1% 9.9%
Percentage

 

 

 

 

 

 

 

 

 

 

 

 

 

73

3.7 Frequency Effects on Material Properties

3. 7. 1 Dielectric Measurements of Uncured DGEBA/DDS

In order to investigate the effects of frequency on dielectric properties of epoxy
resins, experiments were conducted to measure the dielectric properties of uncured
DGEBA/DDS resin with different frequencies at room temperature. DGEBA used in the
study was EPON 828 supplied by Shell Chemical Company. DGEGBA and DDS were
mixed with stoichiometric weight ratio 1 : 0.33. The resin was loaded in a Teflon holder
and the dielectric properties were measured using single mode cavity perturbation method.
An empty Teflon holder was loaded in the cylindrical single mode cavity as a reference
state. Due to the limitation of the experimental setup, measurements were only possible to
be made in the frequency range from 2.2 GHz to 2.7 GHz. The maximum cavity length
prohibits the measurement for frequencies below 2.2 GHz. When frequency goes above
2.7 GHz, resonance peaks of different modes overlap one another to make it impossible to
identify single mode resonance peak. In other words, the Q factor is too low to make
dielectric measurement.

The results are presented in Table 3.4 and in Figures 3.9 and 3.10. No obvious
trends were observed in the experimental data of dielectric constant, neither for dielectric
loss factor. The average of the complex dielectric constants is 3.439 -j0.1328. Since the
frequency range was not broad enough, the fluctuation of the data due to experimental
errors concealed the trends, if any, of the dielectric property change with respect to

frequency.

74

One important factor that affected the accuracy of the dielectric measurement is
the Q factor of the cavity. As indicated by the mode chart, at higher frequencies the
resonance peaks are closer to each other, which leads to lower Q factor of the cavity for a
single mode. The Q factors of the cavity loaded with Teflon holder both with and without
sample are shown in Figure 3.11. The data showed a decrease of the Q factor with the
increase of frequency when the cavity was loaded with empty Teflon holder. However,
there were no apparent trends of Q factor change with fi'equency when the sample was

loaded.

Table 3.4 Dielectric Properties of Uncured DGEBA/DDS

 

f 2.2563 2.3005 2.3502 2.3995 2.4502 2.4999 2.5500 2.6000 2.6498 2.6999

 

5' 3.220 3.647 3.396 3.426 3.460 3.605 3.534 3.680 3.189 3.234

 

 

 

 

 

 

 

 

 

 

 

 

 

0.1384 0.1313 0.1383 0.1286 0.1343 0.1470 0.1352 0.1358 0.1004 0.1387

 

 

 

0.16

0.14 4

1'91
l-H
I'O'l
191
1+1
1.1
101

0.12 +

0.08 ~~

0.06 +

Dielectric Loss Factor

0.04 —~

0.02 ~r

 

 

0 t t t t
2.2 2.3 2.4 2.5 2.6 2.7 2.8
Frequency (GHz)

 

—<>—

 

 

 

Figure 3.9 Dielectric Constant versus Frequency for DGEBA/DDS
75

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

_ I r I
.. I g i
a s =
E
5
g 2.5 “‘7
g 2 -r
G
1.5 a
I i t ‘fi 4 §
2.2 2.3 2.4 2.5 2.6 2.7 2.8
Frequency (GHz)
Figure 3.10 Dielectric Loss Factor versus Frequency for DGEBA/DDS
1.E+04
; I oWithout Sample
1.E+04 + i i i I I I IWith Sample
5
8.E+03 4— i i
‘5 I
E 6.E+O3 ==
6
4.E+O3 I- I
I I 9; I. I i I i i
2.E+03 1—
0.E+00 # t t t t
2.2 2.3 2.4 2.5 2.6 2.7 2.8

Frequency (GHz)

Figure 3.11 Q-Factor versus Frequency

76

 

 

CHAPTER 4

VARIABLE FREQUENCY MODE SWEEPING HEATING

In this chapter, a variable frequency mode sweeping technique is presented, which
was used to unifomrly heat graphite/epoxy composite parts of small size. The concept of
complementary heating is demonstrated using thermal paper images. The cavity loaded
with a unidirectional 2 inch 48-ply square graphite/epoxy composite sample was
characterized. The frequencies and the heating characteristics of the empirical modes were
determined. Four modes with complementary heating preferences were selected for the
mode sweeping heating. The concept of process cycle design is presented for uniform and
fast heating. Both thermal images and temperature profiles were obtained to show the

success of the heating technique.

4.1 Experimental Preparation

Graphite/epoxy prepreg material (I-Iexel" AS4/3 501-6) was used in the heating
experiments. The prepreg laminates were cut into 2 inch by 2 inch squares and laid up uni-
directionally to 48-ply parts. The lay-up procedure for the composite material is given in
Figure 4.1. The weight of each part was about 30 g.

During the experiments, the microwave cavity length was fixed at 15 cm and
coupling probe depth at 20 mm. The composite sample was placed inside a Teflon mold,
the configuration of which is shown in Figure 4.2. The Teflon mold with the sample inside
was placed on the center bottom of the cylindrical single-mode resonant cavity. The cavity

system was not pressurized. The sample was loaded such that the fiber direction of the

77

sample was perpendicular to the microwave coupling probe (Figure 4.3). Since Teflon is

transparent to microwave, the microwave energy only heats the composite material.

 

Teflon Block

 

 

 

Release Film

 

 

Non-porous Film

Porous Film

 

Sample

 

WWW Bleeder Cloth

Porous Film

 

 

Non-porous Film

 

Release Film

 

 

Teflon Block

 

 

Figure 4.1 Composite Material Lay-up Procedure

Thermal paper was used for mapping the heating patterns. The thermal paper was
placed under the sample between the sample and the Teflon material. It changes color
from white to blue at about 85 °C. The dark areas on the thermal paper image indicate
high temperature, and the bright areas indicate lower temperature.

Temperatures were also measured using the measurement probes of a LuxtronTM
F luoroptic Thermometer at four different sites on the sample surface. The temperature
probes were inserted into the microwave cavity and through the top of the Teflon mold.

The probe tips were protected using small glass tubes, which were in direct contact with

78

the sample surface. These temperature probes were composed of optic fibers shielded with
very low loss polymeric materials. Therefore, they do not interfere with the
electromagnetic fields or absorb microwave energy, which assured the accuracy of the
temperature measurement in microwave environment. The dimensions of Teflon molds

used in the microwave heating experiments are given respectively in Appendix A.

 

 

 

 

 

 

 

 

 

 

 

Figure 4.3 Temperature Measurement Locations

4.2 Mode Heating Characteristics

The fi'equencies of the empirical modes were obtained by locating the fiequencies
with locally minimal power reflectance. By tuning the frequency from 2 GHz to 4 GHz,
eight empirical modes were found for the sample-loaded cavity at a fixed cavity length of

15 cm and a fixed coupling depth of 20mm. There is not much change of the locations of

79

the modes with the reloading of the material. Therefore, when a new part is loaded for a
new run these mode locations provide reliable start-points for mode tuning. Temperature
also had minimals effect on the mode frequencies, as shown in Table 4.1, which could be a
result of small sample size.

Typically, the reproducibility of microwave heating profile of conductive fiber
reinforced composites was not very good, due to high sensitivity to sample loading
position, sample size, fiber orientation, fiber content, and fluctuations of microwave
frequency and power. In this study, samples were very carefirlly prepared to increase
reproducibility. Although the exactly same heating profiles were hard to reproduce, the
trends were quite reproducible and were used for characterizing the heating preference of
each mode.

After characterizing the heating pattern for each mode, an optimum heating cycle
can be designed to maximize the uniformity of heating. Computer control algorithm can be
constructed to firrther include the consideration of the heating rates of each mode so as to

optimize the material processing.

Table 4.1 Frequency Shift of Empirical Modes Due to Temperature Change

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Frequency Tuned for Each Mode (GHz)
Mode 0 Mode 1 Mode 2 Mode 3 Mode 4 Mode 5 Mode 6 Mode 7
30°C 2.1477 2.1653 2.1904 2.2446 2.3078 2.3730 2.5555 2.6310
110°C 2.1472 2.1640 2.1918 2.2460 2.3074 2.3722 2.5575 2.6338
f shift -0.0005 -0.0007 0.0014 0.0014 -0.0004 -0.0008 0.0020 0.0028
(GHZ)

 

80

 

The heating characteristics were measured at these frequencies. Thermal images
and temperature profiles of four of these modes (modes 1, 2, 3, and 5) are presented in
Figure 4.4 and Figure 4.5, respectively. As revealed by the thermal images and the
temperature profiles, single mode heating usually results in non-uniform temperature

distribution.

 

 

 

 

 

 

 

 

 

(a) mode 2 (b) mode 3

 

   

 

 

 

 

 

 

(c) mode 4 (1!) mode 6

Figure 4.4 Thermal Paper Images of Four Selected Modes

81

 

 

 

O

..~“~’O
””9”“ J

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E 30 J“ 0 T1
20 r I T2
10 == ‘ T3
0 1 1 1 1 1 X T4
0 5 10 15 20 25 30
Time (nitrate)
(a) Mode l: f= 2.1653 GHz
90
80 “I” “ ‘ . x
A M 0!
70fip g. gg.“§69“"“‘.‘:.!!:!tx
‘. ‘7’00’. .
Q a o
{a 60 A” n 9%..
s 50 {I“Oo°.
a I 08"!”
40 “r ‘60..
g- 0 fl 0 T1
[- 3 I T2
20 “I A T3
1° “" x r4
0 .1 1 + + t
0 5 10 15 20 25
Time (minute)

 

(b) Mode 2: f= 2.1904 GHz

82

 

 

‘3'

 

 

 

 

 

 

 

100
90a
80 . . o . e I»
__ . . . , a I .0
G70" . . I ‘ I ‘ '
4,60 o’ia‘l xxxxx
5 o z I ‘ x x x x
a so = . 3 r x x x g
3.407" 0 g x x x ‘OTII
o 4 x I
E-t 30— ITZ‘
20" AT3
10-- xr4I
0 t I 1 l ”‘i'T‘
0 2 4 6 8 10
Time(mirrute)

 

(c) Mode 3: f= 2.2446 GHz

 

 

 

 

Temperature (degree)

 

 

 

 

0 2 4 6 8 1O 12 14 16 18
Tlmo (minute)

 

((1) Mode 5: f= 2.3730 GHz

Figure 4.5 Single Mode Heating Temperature Profiles of Selected 4 modes

83

4.3 Complementary Heating Concept and Mode Switching Technique

Due to the nature of electromagnetic fields, the field distribution of a single mode
is generally uneven. Since microwave heating effects on non-magnetic material largely
depend on electric field strength, this uneven field distribution often leads to uneven
heating. From the thermal images and temperature profiles of the four selected modes, one
can clearly see the hot spots and cold spots during heating.

Since different modes have different hot spots and cold spots, if the heating
patterns of different modes are superimposed on one another, hot and cold spots tend to
even out. If the hot spots of one mode are cold spots for another mode and vice versa,
these two modes are said to have complementary heating patterns. It is obvious that the
combined heating effects of complementary heating modes produce more even heating
than any single mode heating. Figure 4.6 illustrates the concept of complementary heating

by combining the heating effects of mode 3 and mode 5.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

f = 2.2446 GHz f = 2.3730 GHz combined heating pattern

Figure 4.6 Complementary Heating Using Two Modes

The way to combine the heating effects of different modes is by exciting the modes

in a particular sequence, which is called mode switching. Different design of mis sequence

84

leads to different effectiveness. One example is to excite the modes in a pre-ordered
sequence, which is termed Mode Sweeping technique. More advanced techniques
determine when to select which mode by taking the mode characteristics and on-line
temperature information into consideration. These techniques will be discussed in the

following chapters.

 

 

 

    
 

 

Yes

Heating time

>t1?

 

 
     
 

processing time

>90min?

 

No

 

 

 

 

 

 

Figure 4.7 Mode Sweeping Algorithm

85

4.4 Control Algorithm and Program

The algorithm for mode sweeping heating is very straightforward, which is given in
Figure 4.7. The computer wrote the frequency value to the HP Oscillator in a
predetermined order. The frequency would remain the same for a certain period of time.
Difi‘erent frequencies could have different duration. The software program

(modesweep.vi) was written in LabVIEW, and presented in Appendix D.

4.5 Selective Mode Sweeping Heating Results

These four modes with heating characteristics presented in Figures 4.4 and 4.5
have relatively high heating rates and complementary heating patterns. Therefore they
were selected to heat the sample by discretely sweeping of the mode frequencies. During
the heating, one frequency corresponding to a mode was selected at a time and remained
for some time, then another frequency would be selected. The heating time for each mode
and the order of the modes were programmed. Then the constructed heating cycle was
used to heat the sample until the highest temperature reached 85 °C. Only the frequency
was controlled by computer, by setting the value of the frequency output of the oscillator.
The probe depth was fixed at 24.0 mm during the heating experiments. And the power
was not controlled with a fixed dial knob position. Measurements for all four modes
showed that the power output ranged fiom 23 Watts to 28 Watts.

The first mode sweeping heating experiment let each mode have the same heating
time: 0.1 second. The order of the modes within a cycle was: mode 1, mode 2, mode 3
and mode 5. The thermal image and temperature profile, shown in Figure 4.8 and Figure

4.9 respectively, showed that the middle and lower part of the sample had higher

86

temperature than the upper part. However, the temperature gradients shown by the
temperature profiles were less than 6 °C. Since the thermal image of the first run looks
similar to the heating pattern of mode 1, the residence time of mode 1 should be
decreased. In the heating cycle of the second run, mode 1 had a heating time of 0.5
second in one cycle. The other ones all had 0.1 second heating time. The thermal image
and temperature profile of the second run showed better uniformity of heating (Figure 4.8
and Figure 4.9). The temperature gradients were less than 3 °C when the temperature
reached 85 °C, as shown in Figure 4.9.

Due to the heat loss, the edges of the sample tended to have lower temperature
than that of the center. Both thermal images showed that the center temperature
eventually tended to be higher than other ones when the heating is uniformly initially.
This non-mechanistic temperature gradient increase affects the uniformity of the heating,
and thus the properties of the product. To eliminate the heat loss, the cavity may be
insulated or heated from outside using thermal method to provide a pseudo-insulating

condition.

 

 

 

 

 

 

(a) Mode Sweeping Heating I (b) Mode Sweeping Heating 11

Figure 4.8 Thermal Paper Images of Mode Sweeping Heating

87

 

l r...

V
-"1..I .1': l".

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1f- . .' .‘*
8° -'ri!!:A5“““
7o -_ .i,‘!8!ee
G X...‘.‘A ‘
L 60 4” "’
ii“.
g so —~ “.3
g 40 + "uh.
[- 20 I T2
A T3
10 is x T4
0 1 1 + 1
0 5 10 15 20
Time (minute)
(1!) Mode Sweeping Heatingl
90
80a .ile‘ilii
el"'.’
70 -_ l 4"
6‘ “a.“
a. 60 + N g a !
E 50 ~~ ‘3"
2 40 is...:
g. 30 r- 0 T1
FN+ 'n
A T3
10 “T— x T4
0 1 1 .1
0 5 10 15
Time (minute)
(b) Mode Sweeping Heating 11

Figure 4.9 Mode Sweeping Heating Temperature Profiles

88

 

 

4.6 Summary and Conclusions

The experiments demonstrated that the power source consisting of an oscillator, a
RF plug-in and a TWT worked well as a variable frequency power supply. This variable
frequency microwave system, using a cylindrical cavity as the applicator, was able to
obtained heating modes with a variety of heating patterns when graphite/epoxy material
was loaded. A more uniform heating pattern resulted when two modes with
complementary heating preferences were used alternatively for heating.

A mode sweeping heating technique was designed to take advantage of
complementary heating concept and improve the temperature uniformity of microwave
heating. Mode selection cycles were designed. In each mode selection cycle, a sequence of
modes was used to heat the sample, each mode for a certain period of time. The mode
selection cycle was repeated until the end of processing.

Mode sweeping method was shown to heat the 2" square graphite/epoxy
composite parts not only uniformly, but also efficiently. The efiiciency of mode sweeping
heating lied between the highest one and the lowest one among those of the modes used.
As observed in the experiments, mode sweeping with equal time intervals did not obtain
optimum heating. To get even more uniform heating, the heating characteristics of each
mode should be considered in the optimum heating process design. The heating uniformity
of this method shows the potential to achieve uniform curing of the graphite/epoxy

material of small size.

89

CHAPTER 5

INTELLIGENT VARIABLE FREQUENCY MODE SWITCHING PROCESSING

Intelligent mode switching processing using variable frequency microwaves is
presented in this chapter. The rationale of intelligent mode switching heating is discussed.
Based on the firndamental concepts of the intelligent mode switching method, a process
control system was developed, which includes a mode selection controller, a mode tuning
controller, and a microwave power controller. A simple parabolic power control algorithm
was used to provide stable curing temperature control. Both stepper motor and variable
attenuator were used to microwave power adjustment. The material processing
performances using these devices are compared. Experimental results of square and V-
shaped graphite/epoxy composite parts are presented. The results show the effectiveness

of mode switching heating technique in reducing processing temperature gradients.

5.1 Rationale of Intelligent Mode Switching Heating

Microwave heating is instantaneous, volumetric and non-uniform in nature.
Different modes have different electric field distributions, which are typically non-uniform.
However, uniform heating can be obtained using combination of modes. In a multi-mode
oven, the applicator is over-moded and time averaged uniform heating can be obtained by
frequency sweeping. However the energy efficiency of this practice is very low, since a
large portion of the time microwave energy is not critically coupled into the cavity or the
material. Single-mode cavities can provide uniform heating using mode-switching method,

in which several modes with complementary heating patterns are alternatively excited.

90

With a fixed fiequency microwave power source, mode switching can only be achieved by
mechanically adjusting the volume of the cavity, which is usually accomplished by moving
the sliding top via a stepper motor. This mechanical process slows down the response of
the system to temperature changes. Experiments showed that it took from 30 seconds to
more than one minute for the cavity length to be adjusted with a satisfactory precision.

When a variable frequency power source is available, the heating modes can be
changed by varying the frequency. Frequency control is available as an electronic process,
in the form of digital communication between the computer and the variable frequency
microwave power source. As a result of the instantaneous variable frequency mode
switching, not only the speed of the process but also the controllability of the process is
much improved. The use of variable frequency microwave power source also enable the
control system to periodically tune the mode to make sure that microwave energy is
critically coupled into the material.

With the implementation of computer control system for controlling the heating
modes, the benefits of microwave heating can be fully utilized while achieving uniform
processing. An automated microwave processing system with variable frequency power
source and single mode cavity is presented in this chapter. The control parameters were
microwave frequency and power. Heating modes with certain frequencies were
characterized before processing. The computer chose the frequency that can improve the
temperature uniformity the most by comparing the heating characteristics at that frequency
with current temperature distribution. When the maximum temperature reaches curing
temperature, microwave power was adjusted so as to keep the maximum temperature at

constant.

91

5.2 Process Control System - VFMPCS I

The process control system for intelligent variable frequency microwave
processing consists of a mode tuning controller, a mode selection controller, and a power
controller. Figure 5.1 shows the configuration of the process control system. The
LabVIEW program (VFMPCSI.vi) is documented in Appendix D.

The intelligent mode switching heating requires a mode selection algorithm that
chooses the optimal mode to heat the sample. In addition, during the processing, it is
desirable to periodically tune the frequency or mode so as to minimize the power
reflectance. Moreover, when the temperature is at the curing level, the control system
must be able to keep the curing temperature at the desired level. The maximum of the

surface temperatures was used as the feedback for curing temperature control.

 

 

 

 

 

 

 

 

 

 

 

 

 

T_,C @— ,| P1 l f
ower . .
l d
Controller r ' Cy m ncal
Emodification ' ---------------------
e A' 7 7 .
AT...“ : _
i Mode Selection Mode Frequency
Controller Tuning Controller
........................... Temperature . ........................................ 5
Sensor

 

Figure 5.1 Process Control Diagram for Variable Frequency Mode Switching Heating

92

As seen in the diagram, the curing temperature is given as a set-point for the
control system. The heating characteristics of each mode, including heating rate and
heating pattern, are measured before the curing and stored in a database. When the
heating is started, the temperatures are measured and the data are analyzed. If the
temperature difference at different locations exceeds some value, the mode selection
subprogram starts to search for another mode that has a heating pattern complementary to
the current temperature distribution. The heating rate of the modes is also considered so
as to obtain better energy efficiency as well. Once a mode is selected, the computer will
adjust the microwave frequency and coupling probe depth to the corresponding values.
Around these values, the mode tuning subprogram will tune the frequency and the probe
depth, so as to minimize the reflected power. After the tuning, the frequency and probe
depth values will be used to update the database. This will take into account the changes
of mode locations due to the changes of the temperature and the degree of cure of the
material. Also, during the heating, the heating pattern of the mode will be measured and
the database will be updated. As the temperature reaches the curing temperature, the

input power will be adjusted by a power controller to maintain a constant temperature.

5. 2.1 Mode Tuning Controller

After the computer selects a heating mode, the frequency of this mode (fl) will be
given to the mode tuning controller. The task of the mode tuning program is to minimize
the reflected power around the given fiequency. The procedure is similar to that used in
the mode characterization program. The difference is that the mode tuning controller

searches for the mode in a small range (fl, - Afo , fl, + Afo ). The mode tuning controller

93

increases the frequency fiom (f0 - Afi) with an increment of Af at each time. The input
power is analyzed as a fiinction of the frequency. If the reflected power curve reaches a
minimum, the corresponding frequency will be recorded. If there are more than one
minimum, the one with corresponding frequency closest to f, will be selected. The sample

will then be heated with the selected frequency.

5. 2. 2 Mode Selection Algorithm

Before microwave processing experiments, frequency scan was conducted from 2
GHz to 4 GHz for the cavity loaded with sample. Both input power and reflected power
were measured. Power reflectance curve was obtained by plotting percentage of reflected
power versus frequency. At frequencies with low power reflectance, the sample usually
absorbs the microwave power significantly. Frequencies corresponding to locally minimal
power reflectance are identified as frequencies of potential effective heating modes.
Temperature rises were measured at six locations for each frequency. The heating rates
were obtained by doing a linear fit to the temperature curves during heating-up stage. The
heating rates for each mode were then normalized to between 0 and 5, using the following
formula:

Ri' =(Ri-ijn)/(Rm-Rm;n) x 5, i = 1, .., m (5.1)

Where m is the number of temperatures measured, R; is the heat rate, R'; is the normalized
heating rate, R,m is the maximum heating rate, while Rm is the minimum.

During heating experiments, when the maximum difference between temperatures

exceeds a preset limit (e.g., 10°C), the computer predicts of what the temperature profile

94

would be like if other modes were used. Specifically, temperatures are predicted for each
mode using the normalized heating rates:

T,'=T;+Ri‘><C,i=1,..,m (5.2)
where T,’ is the predicted temperature at location i. C is the time constant that can be
adjusted to change the assumed heating time for the mode used for prediction. The
standard deviation of the predicted temperatures will then be calculated. This will be used
as the measure of temperature uniformity. The higher the standard deviation is, the less
uniform the temperatures are. Therefore, the mode that produces the smallest standard
deviation will be selected as the new heating mode. In essence, the controller picks a mode
that will lead to most uniform temperature distribution after time t.

The larger C in Equation 5.2 is, the firrther in the future the prediction is made. For
example, if the temperatures afier time t are to be predicted, then C should be:
C = Rmm + 5 X t (5.3)
In our case, C had a value of 2, corresponding to a prediction about 30 seconds in
advance. C can be different for heating-up stage and curing stage, because the temperature

dynamics are different.

5. 2.3 Power Control Algorithm
The purpose of the power control program is to maintain the maximum of the

measured temperatures at the set point (the curing temperature of 160 °C). A parabolic

algorithm was designed for this purpose. When the maximum of the temperatures is

between the curing temperature control window, say 157 °C - 160 °C, the power output is

95

parabolically related to the difference between the maximum temperature T and the curing

temperature T:

2
P )x Team—T (5-4)

min T T

Pi = (P
cure — low

max _

where Pi is the input power, Pmax is the set maximum power (100 Watts), Pm,“ is the set
minimum temperature, T is the maximum of the measured temperatures, Tm is the lower
end of the curing temperature control window (157 °C), and Tcure is the curing

temperature.

 

 

100

b) A U! 0\ \l 00 \O
O O O O O O O
1 1 L 1 r I r

201

Input Microwave Power (W)

 

 

 

.'_. ._.
ooo
I

 

l i l l

156 157 158 159 160 161
Temperature (°C)

 

 

 

Figure 5.2 Power Temperature Relationship

If the maximum is higher than the curing temperature, then the microwave power
will be turned off. If the maximum below the control window, then maximum power (100

watts) will be used. A curve representing the relationship is shown in Figure 5.2. The

96

parabolic relationship was used because it is simple and requires no controller tuning. Also
as shown in Figure 5.2, a slight change in temperature would cause more dramatic power
change when the temperature is closer to 160 °C. This is desired because of two reasons.
One is that when the temperature drops below 160 °C, microwave power needs to be
increased quickly to prevent firrther drop of the temperature. The other reason is that
when the temperature gets close to 160 °C, the microwave power needs to decrease

quickly to prevent temperature overshoot.

5.3 Variable Frequency Microwave Processing of Square Graphite/Epoxy

Composite Parts

5. 3. 1 Microwave Power Adjustment Using Stepper Motor

A parabolic power controller was used during the curing stage to keep the
maximum surface temperature at the curing level. The microwave power control was
achieved by adjusting the dial knob on the microwave power source via a stepper motor.

The control algorithm and LabVIEW program are presented in Appendix B.

5. 3.2 Experimental Results and Discussion

The material used was Hexel AS4-3 501/6 graphite/epoxy prepreg. 24-ply
unidirectional 3" by 3" parts were processed. The samples were placed in a Teflon mold
and eight temperatures were measured on the sample surface. The mold was placed on the
bottom of the single mode cavity with cavity length of 15 cm and coupling probe depth of
20 mm. The configurations are the same as in Figure 4.1 and Figure 4.2 in Chapter 4.

Surface temperatures were measured at eight different sites on top of the sample, as

97

shown in Figure 5.3. The sample was loaded into the cavity such that the fiber directions
were perpendicular to the couple probe. During the experiments, the cavity length was
fixed at 15 cm and the couple probe depth fixed at 20 mm.

Fiber Direction

V T

 

 

 

 

 

Coupling Probe Direction

Figure 5.3 Temperature Measurement Locations

 

0.9 "

0.8 ‘

0.71

0.6 ‘

 

 

0.5‘

0.4 "

Preflect/Pinput

0.3 ‘

0.2 1

0.1‘

 

 

 

 

 

0 V V V j' V f fir V V V V V V V V V T V V Y
2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4

Frequency (GHz)

Figure 5.4 Percentage of Reflected Power versus Frequency

98

Before the curing studies, a scan of percentage of reflected power versus
frequency was obtained by sweeping the frequency from 2 to 4 GHz with the sample
inside the cavity (Figure 5.4). The frequencies at troughs with low percentages of reflected
power usually correspond to heating modes, which can heat the sample rather
significantly. Single mode heating experiments were performed at these frequencies.

The maximum of the measured temperatures was controlled as the curing
temperature at curing stage. A description of the control program is given in Appendix D.
Most of the modes demonstrated large thermal gradients, usually more than 40 degrees
when the maximum temperature is at 100 °C. As the heating went on at the curing stage,
the thermal gradients were slightly reduced due to thermal conduction. There was only

one mode that showed relatively uniform heating. The difference of the temperatures was

within 20 °C at the curing stage. However, thermal gradients exceeded 30 0C during the
heating up stage. The heating profiles of a typical non-uniform heating mode and the rare
uniform heating mode are presented in Figures 5.5 and 5.6, respectively. As seen in Figure
5.5, even after 90 minutes of heating, the thermal gradients were still large and the heating
profile was still demonstrative of heating preference. Although the Teflon mold has low
thermal conductivity, the heat loss from the sample to the ambient was significant due to
large temperature difference (120 °C). To compensate for the heat loss, the sample needed
to absorb necessary amount of microwave energy. Since microwave heating was uneven at
that particular mode, the heating preference was preserved in the temperature profile

throughout the heating experiment.

99

Temperature (°C)

 

 

 

 

 

 

 

 

 

 

 

 

Iri-Ir-I'IYVTY‘V’W’V’Y

5 l0 l5 20 25 30 35 40 45 50 55 60 65 70 75 80 IS 90

Time (minute)

Figure 5.5 Single Mode Heating at f=2.5737 GHz

100

 

 

l80

Temperature (°C)

 

 

    

 

‘ f i ' t v

40 50 60 70 80 90 100

Time (minute)

Figure 5.6 Single Mode Heating of 3" Square Sample - f = 3.0818 GHz

Six heating modes were selected for the mode switching heating experiment. The
criteria of selection were the complimentariliness of the temperature profiles and the
heating efficiency. The fiequencies of the modes are given in Table 5.1. The temperature
profiles of the variable frequency mode switching heating are presented in Figure 5.7. As
can be seen, thermal gradients were significantly reduced compared with single mode
heating. The temperatures were controlled to be within a window of 15 degrees
throughout the processing. However, the temperature profiles were less stable due to the
combination of mode switching and power control. The change of modes selected during

the processing is illustrated in Figure 5.8. The six modes were quite evenly used during the

101

processing except for mode 5. The power adjustment is illustrated in Figure 5.9, which is
rather sporadic. As shown in the single mode heating profile, the parabolic power
controller performed reasonably well. However, during the mode switching heating, since
the heating rates of the modes were different, which introduced fluctuations into the
curing temperature control. Since the power control was actuated by mechanical
movement of the stepper motor, the adjustment steps were slow and coarse. The less
satisfactory performance of the power controller for mode switching heating was due to

both the control algorithm and the control actuator.

Table 5.1 Frequencies of the Modes Used in the Mode Switching Heating

 

Modes 0 1 2 3 4 5

 

 

Frequency (GHZ) 2.5772 2.7065 3.1009 3.1643 3.4415 3.6409

 

 

 

 

 

 

 

 

Since the heating characteristics of the modes were determined before the
processing, the success of mode switching technique in alleviating thermal gradients
proved that the heating characteristics are repetitive. The effectiveness of power control

algorithm can be seen from single mode heating profiles.

102

Mode Number

Temperature (°C)

 

 

 

o u v w 1 -

 

Time (minute)

Figure 5.7 Mode Switching Heating of 3" Square Sample

 

 

 

1 9 re as as 41 4s 51 u n so
time (mm)

Figure 5.8 Mode Selection Histogram during Mode Switching Heating

103

 

'r

1N

 

0
OS

Input Power (Watts)

 

 

 

Figure 5.9 Input Power Change during Mode Switching Heating

5.4 Variable Frequency Microwave Processing of V—shaped Graphite/Epoxy

Composite Parts

5. 4.] Microwave Power Adjustment Using Variable Attenuator

In order to improve the power controller performance, a new approach to
microwave power control actuation was developed using a variable attenuator, the
characteristics of which are presented in Chapter 3. The variable attenuator is connected
between the HP Oscillator and the variable fi'equency microwave power amplifier. The
attenuation can be changed via voltage control, by which the microwave power is

controlled.

104

Compared with using a stepper motor to adjust the dial knob on the microwave
power source so as to vary the amplification rate, the use of a variable attenuator provides
much faster microwave power adjustment. An experimental comparison was conducted, in
which the times used by the stepper motor and the variable attenuator to adjust to a new
power level were measured and plotted in Figure 5.10. The initial microwave power was
set at 50 watts, and the desired microwave power was from 10 watts to 90 watts. As
shown in Figure 5.10, the larger the power change, the longer it took the stepper motor to
adjust, while the adjustment time using a variable attenuator stayed pretty much the same.
For example, when the power change was 20 watts, it took the stepper motor 4 seconds

to adjust to the desired power level, and about 1 second using the variable attenuator.

 

 

 

 

 

 

 

 

12 1 Initial Power= 50 watts
A109
'8
8 8 :- oattenuator
5 Isteppermotor
Q I
.E 6—
E-'
a ' '
E 4‘ I I a
.3 u
(i u u
2‘ I . I I
o"’?‘?’?o4=2°:‘t°f
10 20 30 4O 50 6O 70 80 9O
DesiredPower(W)

 

 

Figure 5.10 Comparison of Power Control Performance between Variable Attenuator and

Stepper Motor

105

5. 4. 2 Experimental Results and Discussion

The variable frequency mode switching process control system was used to
process V-shaped graphite/epoxy composite parts. The same material (Hexel AS4-3501/6
continuous graphite/epoxy prepreg) as used for square sample was used. Prepregs were
cut into 3 inch by 3 inch pieces. Twenty-four of them were laid up uni-directionally to give
a sample thickness about 0.17 inches (4.3 mm). The panels were bent to form a V-shape
and placed in a Teflon mold, as shown in Figure 5.11. Temperatures were measured using
fluoroptic temperature probes at six locations, the numbering of which is shown in Figure
5.12. The fluoroptic temperature probes are transparent to microwave. Thus the probes do
not interfere with the electromagnetic fields and do not heat up under the microwave
radiation during the experiments.

The Teflon mold with the sample was loaded into the cavity such that the fiber
direction was perpendicular to the microwave coupling probe. It was determined by
experiments that this was the optimal orientation, with efficient heating and diverse
heating profiles. A power reflectance curve was obtained for loaded cavity and presented
in Figure 5.13. In the next step, the sample was heated using the frequencies at which low
power reflectance was observed. Temperature profiles were measured to obtain the

heating characteristics of each mode.

 

 

 

 

 

 

125°
0 l v '
\‘ - -
.I_ ________ l.
V-shaped sample Teflon mold

Figure 5.11 V-shaped Sample and Teflon Mold Configurations

106

Fiber Direction

‘
r

 

A

 

 

T2 T3 Fnld line
T1 T4

 

 

 

TCoupling Probe

Figure 5.12 Temperature Measurement Configuration

 

 

 

 

 

 

 

 

 

 

 

T— T T l l l r T

2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4
Fiequency(GHz)

 

 

 

 

Figure 5.13 Power Reflectance versus Frequency

During the frequency scan, the temperatures were also monitored (Figure
5.14). With this measurement, one can see whether there are modes that heat
different regions of the sample. As shown in Figure 5.14, the overall heating effect
was quite uniform. Most temperatures had ups and downs during the frequency scan,

which indicated that there were modes with different heating preference. Using this

107

approach, one can monitor temperature changes over the frequency range for
different sample loading positions and determine the optimal position that will give
diverse heating patterns. The more diverse heating patterns are, the more uniform

heating can be achieved by mode switching heating.

 

50

 

eTl
.I'IZ
.13

A
M

    
 

8

U)
M
1

 

 

Tenperature (°C)
w
G

N
M
1

 

 

N
C

 

r f j T T T ‘T

2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4
Frequency (Gill)

 

 

Figure 5.14 Temperature Change during Frequency Scan

The heating characteristics of the heating modes were analyzed and six modes that
had complementary heating patterns were selected for mode switching heating. The
heating profiles of the modes are given in Figure 5.15. As shown in Figure 5.15, single
mode heating was quite uneven. Since the cold spots for some modes were hot spots for
others, uniform heating was obtainable by combining the heating effects of these modes.
Single mode processing was conducted with frequency at 3.6506 GHz (Figure 5.16). This
was the most uniform modes among the available ones. The largest temperature gradient

during processing was about 45°C. The temperature gradient decreased slightly as the

108

as the curing proceeds. This was because of heat conduction from center to cold spots
along the edges.

Mode switching heating was conducted using a curing temperature control
window of 155°C - 160°C, as described in power control algorithm. The temperature
profiles were much more uniform compared with single mode heating, especially during
the heating stage (Figure 5.17). At the curing stage, the edge temperatures were always
lower than the center temperatures due to two reasons. One was the heat loss at the
edges. The other was that the modes that heated the edges preferentially also heated the
center somewhat. Because the maximum temperature must not exceed the curing
temperature, the input power at curing stage was low, this firrther reduced the
effectiveness of edge heating modes. A second mode switching heating experiment was
conducted with a curing temperature control window of 157°C - 160°C. By narrowing the
control window, the average input power increased and the difference between center and
edge temperatures was reduced by about 15% (Figure 5.18).

The heating uniformity during the heating-up stage demonstrated the effectiveness
of mode selection controller. The stability of the maximum temperature during the curing
stage showed that the power controller was successful in maintaining a constant curing
temperature. The curing temperature control was much more stable than using a stepper
motor. The power variation during the processing also showed much smaller fluctuations
than in the case of using a stepper motor. By narrowing the control window, the accuracy
of temperature control can be improved as well as uniformity if heating of the edges was
difficult. On the other hand, over-narrowing of the control window decreases the control

system stability. Since control actions are stronger when control window is smaller,

109

temperature fluctuations would be more pronounced which could result in severe

temperature overshoot.

 

 

180

 

160 -

 

 

 

 

0T1
ITZ
AT3
xT4
xTS
0T6

 

 

Time (ninute)

 

(a) Mode 0: f = 2.1605 GHz

 

Temperature (°C)

 

180

 

 

 

eTl
I12
AT3
xT4
xTS
0T6

 

 

 

 

 

 

F l 7

1 2 3
Time (nI'nute)

 

(b) Mode l: f = 2.7243 GHz

110

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

20 . 1 r r u
0 1 2 3 4 5 6
Time (nI'nute)
(c) Mode 2: f= 3.1610 GHz
180

 

 

 

 

 

 

 

 

0.6 1.1 1.6 2.1 2.6 3.1 3.6
Tim (ninute)

 

((1) Mode 3: f= 3.5420 GHz

111

 

 

1‘80
160 —
14o -

 

Tenmerature (0C)
Os oo 8 :3
O o o o

A
O

 

 

 

N
O

 

 

, “"593?!" -

(e) Mode 4: f= 3.6506 GHz

 

Terrperature (°C)
00 8
o c

«13¢
00

 

 

 

20 1 l l .
0 1 2 3 4 5
’ Time (ninutc)

 

 

(1) Mode 5: r= 3.7911 GHz

Figure 5.15 Temperature Profiles of Single Mode Heating

112

e-
- fe-
v
e),
t
'.‘4 ‘
U‘.
h
4.
e
A
r
7’...

(a

 

 

 

Time (“0

 

 

 

 

0 l l l’ T F _—
0 10 20 30 40 50 60
Time (ninute)

 

Figure 5.16 Single Mode Heating at f = 3.6506 GHz

 

 

 

 

 

 

 

 

 

 

180
160 4
140 1 ' - ‘ ' ‘ ‘ '
E 120 -
5 100 . erfl
a so - .3 " T2
E- 60 fl 9‘ . T3 .
g 40 l'. x T4 1
at T5
20 “ . T6
0 1 I 1 1 T
0 10 20 30 4O 50 60
Tim (ninute)
(a) Temperature Profiles

113

 

’e’

 

 

3 ——n——¢-Ao—u—~+«N ~4—a ore—9&7- 4». H—+-+Q—oeflewoo

 

2 wowow

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1
0 4" r r I I T
0 10 20 30 40 50 60
Time (minute)
(b) Mode Selection Histogram
90 e r .
80 ~ . Input Power
70 - . . Rekcted Power
60 ~
€50 1.
3 4o - . - .‘
3O _. e . ‘...‘ h. . .‘e.. . . 9. . . . e
20 _ a. a . .u 0‘ .‘.e.“4 a. .5... .M.. .~. ‘3‘...
10 —« ‘
o e: -1 -,. 4e see -1 T -
0 10 20 30 40 50 60
Time (ninute)

 

 

 

(c) Power Variation During the Processing

Figure 5.17 Mode Switching Heating 1 with Curing Temperature Control Window:

155°C - 160°C
114

Help

 

 

180

 

 

   

 

 

 

 

 

 

 

160 ~
140 .
3 120 ~
5 100 - 0 T1
i so ~ '3
A
g 60 " l x T4
40 I a: T5
20 i e T6
0 . 1 . . .
o 10 20 3o 40 so 60
Thu (ninute)
(a) Temperature Profiles

 

 

3 AMo—Hmwe £-—-H—--i +& QHH--H——+-—H-~

 

 

 

 

 

 

 

 

2 2 «0 e
l e e
0 M r T T T I
0 10 20 30 40 50 60
Tim (ninute)
(b) Mode Selection Histogram

115

 

‘1'

. ‘ v‘i‘

.| ;‘I€.P. no,“ I .

..-.‘—

54". “‘4 ".

A,“

.a l "'"."¢“'MJT dirt" "‘ w ..

 

 

80*
70*

€60“

5.501

30~
20‘

£403 ‘

10~

 

 

- Irput Power
. Rekcted Power

 

 

 

 

 

 

10 20 30 40 50 60
Tim (ninute)

 

 

(c) Power Variation During Processing

Figure 5.18 Mode Switching Heating 2 with Curing Temperature Control Window:

157°C - 160°C

5.5 Summary and Conclusions

In variable frequency mode switching heating, processing cycles were designed

based on the mode heating patterns only. However, during the processing of composites,

the temperature distribution varies. Different temperature distributions call for mode with

different heating preferences. An intelligent variable frequency mode switching technique

was designed and developed in LabVIEW (VFMPCSI) to match mode characteristics

with sample temperature distribution when selecting the heating mode.

116

The control system predicts the temperatures using the mode heating rates and the
measured temperature distribution. The mode that will result in the smallest standard
deviation of the predicted temperatures is selected for heating. A stepper motor was used
to adjust the manual knob on the microwave power source. A parabolic power control
algorithm was designed for simple microwave power control that did not require
controller tuning. The power control algorithm was proved to be effective and stable.
However fluctuations occurred during mode switching heating due to the different mode
characteristics and the coarse and slow adjustment of the stepper motor for microwave
power control. 24-ply unidirectional 3" by 3" graphite/epoxy composite parts were
successfully processed using the intelligent variable frequency mode switching technique.
Heating experiments proved that the heating characteristics of each mode was repetitive as
long as the sample was loaded in the same way.

Variable frequency mode switching technique was also employed to significantly
improve the heating uniformity of microwave processing of V-shaped graphite/epoxy
panels. A different microwave power control hardware - variable attenuator, was used.
The power adjustment time for the variable attenuator was much smaller than that of the
stepper motor. The adjustment time using variable attenuator was less than 1 second,
while the adjusment time for stepper motor ranged from 1 second to more than 10

seconds depending on the power change size.

117

CHAPTER 6
VARIABLE FREQUENCY MICROWAVE PROCESSING OF COMPLEX

SHAPE COMPOSITE PARTS WITH ON-LINE MODE UPDATING

In this chapter, an on-line mode heating characteristics updating algorithm is
presented. On—line mode updating ensures that the mode selection controller uses the
accurate information about the mode heating characteristics when selecting the optimal
mode to achieve uniform heating. The mode selection controller is upgraded to make
more accurate comparison of the abilities of the heating modes to alleviate temperature
gradients. A multi-staged PID controller was designed to provide more stable and precise
curing temperature control. Complex shaped graphite/epoxy composite parts were

processed using the process control system with on-line mode updating (VFMPC S 11).

6.1 On-line Updating of Mode Heating Characteristics

For mode switching heating technique, the computer always tries to find a mode,
when necessary, that will heat the sample most uniformly. To do so, the computer needs
the information about the heating characteristics, i.e. heating preferences, so that it can
predict how the temperature distribution will change if the mode is used to heat the
sample. In order for the mode selection controller to be effective, the information about
the mode heating characteristics must be accurate.

Mode heating preferences are characterized before processing experiments, as
discussed in previous chapters. The heating rates at the temperature measurement sites
during the heating-up stage are computed and stored in the computer. This information
can be used for subsequent processing experiments as long as the sample material,

118

configuration and loading position, and the cavity settings are the same. Obviously, these
measured heating rates are only approximate representation of the mode heating
preference, which varies with different batches of samples. In addition, mode heating
characteristics or preferences change during the processing, not only because of the
change of temperature distribution, but also because of material property changes during
the processing, such as extent of cure, dielectric properties, and thermal properties. It is
impossible to characterize exactly how each mode heats at certain stage of the processing.
However, by on-line measuring the mode heating characteristics, the accuracy of mode
heating preference information can be improved. This updating of mode heating
characteristics can be carried out by computing the heating rates at each temperature

measurement site for the mode that is being used.

6.2 Process Control System - VFMPCS II

The VFMPCS II process control system is consisted of a mode tuning controller, a
mode selection controller, a multi-staged PID microwave power controller, and an on-line
mode characteristics updating controller. The algorithms of these controllers are discussed
in the following sections. The LabVIEW program (VFMPCS.vi) for the control system is

documented in Appendix D.

6. 2.1 Mode Tuning Controller

The mode tuning controller has the same algorithm as the one described in Chapter
5. The purpose of mode tuning is to minimize the power reflectance around the given
mode frequency. Mode frequencies were determined by locating the valleys of the power

reflectance versus frequency curve. The shape of the power reflectance versus frequency

119

curve depends on the material dielectric properties. Since material dielectric properties
change during processing, so do the valleys of the power reflectance versus frequency
curve. Therefore, mode tuning can track the shifi of the mode frequencies and make sure
that the coupling between the microwave power and the sample is maximum. Mode tuning
is carried out every time the heating mode is changed or the fiequency is changed. After

mode tuning, the tuned frequency will be used as the new fiequency for the selected mode.

6. 2. 2 Mode Selection Controller

As similar in the mode selection controller in VFMPCS I described in Chapter 5,
the standard deviation of the temperatures is used as the measure of temperature
uniformity. The smaller the standard deviation, the more uniform the temperature
distribution. The heating modes are characterized by the heating rates at the temperature
measurement sites: 617‘, /dt, dT6/dt; with [dT/dt] = °C/second. For mode m (m = O to
maximum number of modes), the heating characteristics are: (dT, /dt)m, ..., (dT6 /dt),.,. The
temperatures are measured during the processing: T 1 , , T 6. The process controller
assumes that mode m is used for heating and predicts what the temperature distribution
will be after a series of time intervals: T 1 + (dT, /dt),,,x n x At , , T 6 + (d7; /dr),,,x n x
At; where At is the time interval (e. g. 6 seconds), and n = 1,2,. . ., are the prediction steps
(e. g. n = 1,2,. . . , 30). In other words, the process controller predicts how the mode will
heat the sample every At (seconds), up to At x max{n} (seconds) after the mode is used.

The process controller then computes the standard deviation of the temperatures at

each prediction step:

120

 

2

 

dT.
T.+ —’ xnxAt —[T+[fl]xnxAt
' dt dt

0' = " (6.1)

 

 

 

 

where [T +[fl]x nx At] is the average of the predicted temperatures. The minimum of
dt

all the standard deviations is then determined - 0' m , which is indicative of the ability of

the mode to alleviate or aggravate the temperature gradients. After the same computation
is carried out for each mode or, 0"" of all the modes are compared (m = 0, 1, ...). The

mode with the smallest 0"" is considered as the best mode that will achieve most uniform

heating after certain amount of time. Suppose 0k 2 min b’" }, then mode k is selected
for heating the sample.

6. 2.3 Multi-staged PID Microwave Power Controller

A multi-staged PID controller was developed for variable frequency microwave
processing in order to achieve optimal processing. The processing of materials can be
viewed as composed of different stages. The first stage is the beginning of the processing,
when the microwave input power needs to be at maximum so as to heat the sample as fast
as possible. When the sample temperature approaches the curing temperature, the control
system should be able to slow down the heating appropriately to prevent temperature
overshoot. At the curing stage, the microwave power should be just enough to maintain
the curing temperature. Accordingly, the power controller needs to have different
parameters at different stages. Microwave Processing of composite parts was divided into

four stages as shown in Figure 6.1.

121

Over-heating stage:

 

 

 

 

T Pi = 0
Curing 161 °C
temperature: m 160 °C ¢ Curmg: PID 2
t Curing: PID 1
150 °C
Heating-up stage:
Pi = PM = 90 W
Room
temperature:
30 °C

Figure 6.1 Multi-Staged PID Control

The maximum measured temperature is used to identify which stage the processing
is in. The first stage is the heating-up stage, when the maximum surface temperature is
well below the curing temperature and the maximum microwave is used to heat the sample
as fast as possible. The maximum microwave power used in the experiments was 90 watts.
The curing stage is defined as the temperature range from 10 °C below the curing
temperature to 1 °C above the curing temperature. The curing stage is further divided into
two stages. The curing stage I is between 150 °C and 159 °C, while the curing stage II is
between 159 °C and 161 °C. In the curing stages, PID algorithm is used for microwave
power control. Two sets of PID control parameters correspond to the two curing stages,
(Kpl, T51, TH) and (K92, T12, Td2). When the maximum surface temperature goes above
161 °C, the microwave power will be shut down.

Two different sets of PID parameters are used because in the two curing stages the

temperature increasing rates are different, so are the process dynamics. In curing stage I,

122

the heating rates and input microwave are still high, thus quick response of the controller
is needed. In curing stage H, the temperatures are stabilizing, especially the controlled
temperature - maximum measured temperature. The input microwave power also
approaches to a steady level. Therefore, the control response need not be as quick.
However, control accuracy becomes a more important issue. Generally speaking, Kpl >
Kp2, T,1 > T2, and le < Td2. Experiments were conducted to determine optimal values

of the control parameters. The following are used for the PID control in the experiments:

Kpl = 50, Til = 1000, le = 10 (6.2)

K,2 = 20, T2 = 50, m = 30 (6.3)

6. 2. 4 0n-line Mode Characteristics Updating Controller

On-line mode characteristics updating is the measurement of heating rates for the
modes during processing. It is necessary for mode selection controller to be successful. In
order to filter out the temperature measurement errors, the controller computes the
heating rates after the selected mode heats the sample for a certain period of time. Twenty
seconds was used in the experiments. The controller throws away the first few
temperature data, since there is time delay for the heating effects of the selected mode to
show in the measured temperatures. In the processing experiments, the controller uses the
temperature data during the last 15 seconds to compute the heating rates using linear fit.

The heating rates are then stored as the heating characteristics of the particular mode.

123

6.3 Variable Frequency Microwave Processing of V-shaped Graphite/Epoxy

Composite Parts with On-line Mode Updating

The same material, sample configuration, and loading position as the V-shaped
part processing in Chapter 5 were used. The temperature measurement configuration is

different from that described in Chapter 5, as shown in Figure 6.2.

fiber direction

 

 

fold line
T3 T4 ‘/

 

 

 

TCoupling Probe

Figure 6.2 Temperature Measurement Configuration

6. 3. 1 Heating Modes and Their Characteristics

The same heating modes were used for the experiments, as in the experiments
described in Chapter 5. The heating characteristics are described in Chapter 5.
Complementary heating patterns can be found in the heating characteristics of these

modes.

6. 3.2 Mode Switching Heating Results and Discussion

The experimental results of variable microwave mode switching processing of V-
shaped graphite/epoxy composite parts using VFMPCS H are presented in Figure 6.3. As

shown by the temperature profiles, the temperature gradients are significantly reduced

124

throughout the processing experiment. The final maximum temperature difference was
only 15 °C. During the curing stage, T3 and T4 remained high because they were close to
the center of the sample. Usually, the center of the sample would have higher temperature
during the curing stage because of the temperature gradient created by the heat loss from
the sample to its surroundings.

The microwave power control stabilized the maximum temperature at the curing
temperature rather fast without much temperature overshoot. The temperature was less
than 1 °C. At the beginning of the curing stage, there were some fluctuations of the
temperatures. Those were due to the change of heating modes and the fact that microwave
power level was not stabilized yet. The performance of the power controller was stable
and accurate at the curing stage. The average microwave power kept decreasing, as seen
in Figure 6.3 (c), and approached to a steady value at about 20 watts at the end of
processing. The decrease of average input microwave power could be a result of the
increase of ambient temperature, which reduced the heat loss fiom the sample.

From Figure 6.3 (a), it is shown that all modes were utilized at the heating-up
stage. However, only three modes - 3, 4, and 5, were used at the curing stage. Modes 4
and 5 saw the most action, which indicates that modes 4 and 5 heated the edges of the
sample more.

Mode sweeping heating was also conducted for V-shaped graphite/epoxy
composite samples, following the procedure described in Chapter 4. The heating times in a
cycle for mode 0 and 1 were 10 seconds, and 20 seconds for the rest of the modes. Modes
0 and 1 had less heating times because they had slower heating rates or lower energy

efficiency. The results are presented in Figure 6.4. The temperature profiles in Figure 6.4

125

(a) shows that T; and T4 are again high at the curing stage. The temperature gradients
actually increased at the beginning of the curing stage. The reason was the heat loss from
the sample to its surroundings. As the heating progressed, the ambient temperature
increased and temperature gradients decreased. The power variation during the processing
is presented in Figure 6.4 (b). The microwave power level was more sporadic as in mode
switching heating because of the frequent change of modes. Each time the mode (i.e. the
frequency) is changed, the microwave power had to be tuned down first in order to
prevent a microwave power source fault. Therefore, power adjustment was very frequent.
As in the mode switching heating experiments, the same trend was evident that the
average microwave power was steadily decreasing as the heating progressed.

For comparison, the temperature profiles of a single mode heating experiment
were also presented in Figure 6.5. The maximum temperature difference and standard
deviation of temperatures are plotted versus time in Figure 6.6, for these three processing
techniques. The significant improvement in temperature uniformity using intelligent mode
switching heating is obvious. Both the maximum temperature difference and standard
deviation of temperatures were small and stable in the case of variable frequency mode
switching heating. The averages of maximum temperature difference and standard
deviation of the temperatures are listed in Table 6.1. The maximum temperature difference
seems to follow the same trend very closely as the standard deviation. Therefore, similar
results could result if the maximum temperature difference were used as the measure of

temperature uniformity.

126

Table 6.1 Average Maximum Temperature and Standard Deviation at Curing Stage

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Average Intelligent Switching Mode Sweeping Single Mode Heating
(ADM 13.69 °C 30.07 °C 52.15 °C
0(T) 5.51 °C 12.65 °C 21.64 °C
180
16° 4' “7 4" ' .1... ”Timid f"; -- -' . .. IT‘Z‘J... ”:7"
140 - : f
j
E 120 '
a 100 r”— _____ Tl
% 80 T. _— T2 r
E- -——- T3
60 : —— T4 2
40 1. _ —— T5 5
-——- T6
20 1 .
0 30 60 90
Time (minute)
(a) Temperature Profiles

127

 

 

 

 

eh

L0
0
11
II
II
0
0
0
ll
0
l
0
l1
0
11
o
o
o
o
1

 

1
11

Heating Mode Selection

1
1
1
1
1
1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0 = I 1
0 30 60 90
Time (ninute)
(b) Mode Selection Histogram
100 , "W
I” . | - Input Power
80 . -’ :j 7 . Rech Power ifl__<
. .° - a. ‘ #
60
9?.
0
>
E 40
E
.9
E
20
0 -

 

 

 

Time (ninute)

 

 

 

(c) Power Variation during Processing

Figure 6.3 Intelligent Variable Frequency Mode Switching Heating of V-shape
Graphite/Epoxy Composite with Orr-line Mode Updating

128

1" \. 'Vl, r. '-en' 1 - . '
.5 e! ‘ ‘ "d v I.
I O V
I
\
i
a
't
.L w l
,~ in. " h
. ‘ I,
.
.
‘I a t
i

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I60
140
E 120
E Mode Heating
E 100 Time (s) “T1
a. 10 —
Q 80 10 _T3
60 20
20
40 20
20 20
Time (minute)
(11) Temperature Profiles
100
0 Input Power
80 1a! - Reflected Power
g t .
.. ‘ . 3
E 60 ‘ .es.
2 O
E 40 -
.2
2
20 -
0 ..
0 30 60 90
Time (mute)

  

 

 

(h) Power Variation During Processing

Figure 6.4 Mode Sweeping Heating of V-shaped Graphite/Epoxy Composite

129

 

 

.
an

 

 

 

Temperature (°C)
8 8 8

8 S

 

Time (minute)

 

Figure 6.5 Single Mode Heating at f = 2.1605 GHz for V-shaped
Graphite/Epoxy Composite

 

 

Comparison of Maximum Temperature Difference

—— Intelligent Switching
— Mode Sweeping

(°C) _
888883888

y—s
0

Maximum Temperature Difference

 

O

0 30 60
Time (minute)

 

(a) Maximum Temperature difference Comparison

130

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Comparison of Standard Deviation of Temperatures T
45
40 -- Intelligent Mode Switching _L
-—— Mode Sweeping
E fl —- Single Mode Heating fl
8
.8
p
0
G
E
5
m
L Time (minute)

 

 

(b) Standard Deviation Comparison

Figure 6.6 Comparison of Temperture Uniformity for Single Mode Heating, Mode
Sweeping, and Intelligent Mode Switching Heating of V-shaped Graphite/Epoxy

6.4 Variable Frequency Microwave Processing of Tri-planar Graphite/Epoxy

Composites with On-line Mode Updating

In order to further test the performance of the variable frequency mode switching
processing technique and the process control system, a more complexly shaped geometry
was considered. Twenty-four-ply of uni-directionally laid-up graphite/epoxy prepregs
(Hexel AS4-3501/6) were bent into tri-planar shape as shown in Figure 6.7. The sample
was loaded into a Teflon mold and then into the microwave cavity. The fiber direction of
the sample was perpendicular to the coupling probe. Temperatures were measured at six

different sites on the sample surface, as shown in Figure 6.8. The lower edge of the

131

tri—planar sample was placed near the coupling probe. During the experiments, the cavity

length was fixed at 15 cm, and the coupling probe depth at 20 mm.

\\

Tri-planar sample
Figure 6.7 Configuration of Tri-planar Graphite/Epoxy Samples

fiber direction

 

Higher edge T1 T2

 

T3 T. > fold lines

 

Lower edge T5 T6

 

 

 

TCoupling Probe

Figure 6.8 Temperature Measurement Configuration of Tri-planar Samples

6. 4. 1 Heating Modes and Heating Characteristics

In order to determine the frequencies of empirical heating modes, power
reflectance was measured while the frequency was swept fi'om 2 GHz to 4 GHz. The
power reflectance versus frequency curve is shown in Figure 6. 9. The frequencies with
power reflectance less than 0.1 were used individually to heat the sample and only the
ones with considerable heating effects were regarded as heating mode fiequencies. Figure
6.10 shows the temperature change during the frequency scan. The temperature profile

132

indicate the difficulty to obtain uniform heating because the heating preference remained
almost the same throughout the frequency spectrum.

Six modes were selected for the experiements based on the heating characteristics
and heating effectiveness. The heating characteristics of these modes are given in Figure
6.11, corresponding to the frequencies. Only the heating characteristics at the heating-up
stage were measured, since on-line mode updating was going to be used in the
experiments to provide accurate heating charateristics information. As shown in Figure
6.11, all the modes heated the higher edge (T1 and T2) of the sample preferentially except

mode 0, which heated the lower edge (T5 and T6) preferentially.

 

 

.o .O .o .o
0‘ \l W \O '—‘
1 1 1 1

   
 

 

.9
u:

 

 

 

 

Preflected/Plnput
.9
.b
I

 

 

 

 

 

.0
w
1

 

 

 

.0
N
1

 

 

.0
l—i
I

 

 

 

 

 

 

o
4
—r
—(

I l l l l l

2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4
Frequency(GHz)

 

 

 

Figure 6.9 Power Reflectance versus Frequency Curve for a Tri-planar Sample

133

 

 

 

 

Temperature (°C)

 

 

 

201 l 1 1 l 1 l 1 l
2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4

Frequency (GHz)

 

Figure 6.10 Temperature Change during Frequency Scan

 

Temperature (° C)

 

 

   
   
 

 

 

 

160
140 {QT—i

1 m1
120 ' 1 T31
100 - l ' “l

1 x T51
80 ‘ 1 “11,7391
60 —
4o-
20 1 1 1 1

o 1 2 3 4 s

Time (minute)

 

 

(:1) Mode 0: f= 2.1501 GHz

134

..I
-|‘.
:_.'

1
.r
'.L’o
.e ..

~-e

 

 

 

§

5

 

pd

8
l

:3

1

fl

 

 

Temperature (°C)

 

 

 

 

 

Time (minute)

 

(b) Mode 1: r= 2.3019 GHz

 

 

 

 

§

 

"a“

g

y...
1

 

 

 

Temperature (°C)

 

 
  

 

 

Time (minute)

(c) Mode 2: r= 3.6692 GHz

135

 

 

 

 

 

160

140 .

Temperature (°C)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

20 , T T r
0 0.5 l 1.5 2 2.5
(d) MOde 3: f: 3.7104 GHZ
T 160
140 ~ _.__, T1
T2
1 0 4
G 2 T3
a... . T4
5 10° ‘ . T, “W
E ‘ fl ppppp ”M
E 80~ T6 W P . W
0 ,,,,, rhea—..- e-u- t:£::$’ ”’-
E" 60 4 WWW-My,” M
40 . .2...»-
20 f I f T 1 1
0 0.5 l 1.5 2 2.5 3 3.5

 

(9') M0“ 4= f = 3.7472 GHz

136

 

up.

 

 

 

 

 

 

 

 

 

 

160
140 . “*7 T1
, T2
120 « T3
6
L I T4
2 - ,.
2' so , T6
0 ' r ..--—""*
E" 60 -~ ~
4o 1
20 1 T f T I T T
o 0.5 1 1.5 2 2.5 3 3.5 4
Time (minute)

 

 

(1) Mode 5: r= 3.8356 GHz

Figure 6.11 Mode Heating Characteristics

Figure 6.12 shows the temperature profiles for the processing of a tri-planar
graphite/epoxy composite sample at f = 3.8326 6112. The heating was relative uniform at
the heating-up stage, compared with general single mode heating. However, the

temperature gradient increased at the curing stage and the maximum temperature

difference stayed at about 30 °C.

137

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

20 1 1
0 30 60 90
Time (nrinute)

 

 

 

 

Figure 6.12 Single Mode Heating Profile at f = 3.8326 GHz

6.4.2 Intelligent Mode Switching Heating Results and Discussion

The heating results using variable frequency microwave mode switching
technique and VFMPCS II are presented in Figure 6.13. The temperature profiles show
that the heating was quite uniform with the final maximum temperatme difference of
about 15 °C. The power controller performed well and the maximum temperature (T4)
was stabilized within 160 °C i 0.5 °C quickly after it reached curing stage. The
temperature overshoot was less than 1 °C. Other than the fluctuations at the beginning of
the curing stage, the temperatures remained stable with edge temperatures lower than the
center temperatures. The microwave power decreased as the processing time increased

and approached to a steady level around 25 watts.

138

 

Figure 6.13 (b) shows the mode selected for heating versus processing time. For this
particular experiment, only three modes were actively used for heating. Modes 0, 1, and 5
were used at the heating-up stage. Both modes 1 and 5 were used at the beginning of the
curing stage, after which only mode 1 was used for heating. Apparently, during the
processing, the controller decided that mode 1 had the heating characteristics that would
alleviate the temperature gradients the most. Although mode heating characteristics change
during the processing, the initial heating characterization in Figure 6.11 (b) is indicative of
the heating preference of mode 1. Therefore, the mode selection by the controller was
reasonable, since at the curing stage T1 and T2 were the lowest temperatures and mode 1
heated T1 and T2 preferentially.

Mode sweeping heating was also conducted using the same six modes. The method
was described in Chapter 4. Each mode was assigned a heating time of 10 seconds and the
modes were used sequentially. The heating results are given in Figure 6.14. As shown in
Figure 6.14, initially the center temperatures (T3 and T4) were the lowest, which is
predictable since Figure 6.11 showed that all of these six modes heated the edges
preferentially except mode 0. The overall heating effect would obviously show preference at
the edges since all the modes have equal heating times. As the processing progressed, the
center temperatures increased more rapidly and became the highest temperatures. This is
due to the heat loss from the sample to its surroundings around the edges. Just as in the
intelligent modes switching experiment, T1 and T2 became the lowest temperatures.

The comparison of temperature uniformity of single mode heating, mode sweeping
heating, and intelligent modes switching heating is presented in Figure 6.15. The intelligent

mode switching heating with mode updating proves to be far more superior than single

139

mode heating and mode sweeping heating in terms of achieve temperature uniformity. The

average maximum temperature difference and average standard deviation of temperatures at

curing stage are listed in Table 6.2.

Table 6.2 Average Maximum Temperature and Standard Deviation at Curing Stage

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Average Intelligent Switching Mode Sweeping Single Mode Heating
(AT)um 13.52 °C 15.83 °C 26.83 °C
O'(T) 5.39 °C 7.29 °C 11.17 °C
180
160 -r ‘4 — v -- ~ _: - ' .1.
140 _—~ - Aw A _ “fl “—‘2—
8 120
i
g- 100 -----—T1
3 so . “—13
[2 —-—-—T3
60 7 T4
———T5
40 —-—-T6
20 r 1
0 30 60 90
Time (minutes)

 

 

 

(:1) Temperature Profiles

140

 

 

 

 

 

 

 

 

 

 

0 30 60 90
Time (minute)

 

 

 

(b) Mode Selection Diagram

 

 

 

 

90

' - Input Power
. - Reflected Power

05
O
1

Microwave Power (W)
b)
o

 

 

 

 

Tin: (nitrate)

 

 

 

(c) Power Variation during Processing

Figure 6.13 Intelligent Mode Switching Heating of Tri-planar Graphite/Epoxy

141

","

3:092. "

."

’ "l.

{5? r

on’

a"!

a

 

 

Temperature (°C)

 

0 30 60
Time (minute)

 

(11) Temperature Profiles

 

Microwave Power (W)

N
O

 

U) LII G \l W
O

I—
O

 

0

Time (minute)

 

(b) Power Variation during Processing

Figure 6.14 Mode Sweeping Heating of Tri-planar Graphite/Epoxy

142

 

 

.
e
.
i I l . -
a. 4 v
e. ‘ ’1
5‘ 51"» _s‘s".
. ‘ .
“Us“ . . g .fil‘ ”s. . '
I d
. .
.

 

Comparison of Maximum Temperature Difference

——- Intelligent Mode
— Mode Sweeping

1 f: $33E‘_°i°_”f§‘hli,._.,

Maximum Temperatrue Difference
(°C)

 

 

Time (minute)

 

 

(11) Comparison of Maximum Temperature Difference

 

1 Comparison of Standard Deviation ochmperatures

  
 
 

 

—-— Intelligent Switching
— Mode Sweeping
— Single Mode Heating

 

 

  

 

 

Standard Deviation of
Temperatures (° C)

 

 

 

 

 

 

 

Time (minute)

(b) Comparison of Standard Deviation

Figure 6.15 Temperature Uniformity Comparison of Single Mode heating, Mode Sweeping
Heating, and Intelligent Mode Switching Heating of Tri-planar Graphite/Epoxy

143

\r'

.v‘

r

6.5 Summary and Conclusions

The process control system VFMPCS H has been added the capability of on-line
updating the mode heating characteristics. Mode heating characteristics change during the
processing due to material property changes and temperature change. This feature enables
the control system to obtain accurate heating characteristics and ensure the effectiveness
of the mode selection controller. On-line mode characteristics updating is especially
important for complexly shaped composite parts since the processing conditions vary in a
larger degree.

A mode selection controller with a series of prediction steps was designed. Instead
of comparing the heating uniformity at some point in the fiiture, the controller predicts the
temperatures at a number of points in the time after. Only by computing the predicted
temperatures at different points of time, can the potential of each mode to alleviate the
temperature gradient be accurately predicted. A multi-staged PID controller was designed
to take advantage of the quick response of variable attenuator. The processing of
composite materials was divided into four stages, and power control strategies were
designed accordingly.

Experiments were conducted for both V-shaped and tri-planar graphite/epoxy
composite parts. Results showed significant improvement of heating uniformity over single
mode heating and mode sweeping heating. The mode selection controller and the on-line
mode characteristics updating controller proved to be effective and accurate, and
improved the robustness of the process control system. The power controller performed

well and provided quick, stable, and accurate curing temperature control.

144

CHAPTER 7

. SUMMARY AND CONCLUSIONS

This work was intended to accomplish two main objectives. One is the
development of an automated microwave processing system that employs variable
frequency technology. The other is the development of a process control system for
uniform processing of polymer composites using variable frequency microwave energy.

Microwave processing of materials, especially polymers and polymer composites,
has proved in the past to be advantageous over thermal processing approach. Previous
research has demonstrated benefits including faster heating, increased reaction rates,
enhanced glass transition temperatures, improved conductive fiber matrix adhesion, and
better mechanical properties of the products. However, these advantages have not been
exploited at industrial scale because of the difliculty of developing a microwave processing
system easy to operate while providing desired and consistent performance. In particular,
lack of process automation and inability to provide uniform processing are the major
obstacles. Good process control systems can simply the operation of the microwave
processing system and ensure the consistency of good performance. The use of variable
frequency technology in microwave processing eases the task of process automation
because same effects can be achieved by electronically adjusting frequency rather than
mechanically changing the microwave applicator dimension(s).

In this research, a variable frequency microwave processing system has been
developed and automated. Both frequency and power of the microwave energy source

were controllable by the computer. Processing parameters were measurable by computer

145

through a data acquisition system. A process control system was developed to achieve
uniform processing of polymer composites while maintaining a stable and constant curing

temperature. The results and conclusions are summarized as follows.

7.1 Development of an Automated Variable Frequency Microwave Processing

System

A variable frequency microwave processing system was developed based
on the configuration of the fixed frequency microwave processing system. The major
difference was the microwave power source. The variable frequency microwave power
source consisted of an HP oscillator as the signal generator and a TWT amplifier. The
oscillator frequency output range was from 1.7 GHz to 4.3 GHz. However, only the
frequency range of 2 GHz to 4 GHz was used in the processing experiments for a power
level high enough for effective heating. Other microwave circuit components and the
circuit configuration were the same as the fixed frequency microwave system. However
the microwave components were required to be operational from 2 GHz to 4 GHz. The

variable frequency microwave processing system was automated and characterized.

7. 1.1 Automation of the Microwave Processing System

The automation of the microwave processing system mostly involved achieving
computer control of the microwave frequency and power. The fiequency of the
microwave power source was controlled through the GPIB interface between the
computer and the HP Oscillator. The computer could write the desired frequency directly
to the Oscillator through GPIB. A variable attenuator was used for the computer control

of microwave power. The attenuation was controllable via the control voltage. The

146

variable attenuator was connected between the HP Oscillator and the TWT amplifier
because of the maximum input power to the attenuator was 30 dBm.

Measured processing parameters included input microwave power, reflected
microwave power, and the surface temperatures of the material being processed.
Microwave power was measured by power meters with analog outputs. Temperatures
were measured using Luxtron fluoroptic thermometer with analog outputs. The analog
signals were obtained by the computer through a National InstrumentsTM data acquisition
board. The process control system analyzes the measured processing parameters and

adjusts the microwave frequency and power accordingly to achieve the processing goals.

7. 1. 2 Characterization of the Variable Frequency Microwave Processing System

The characteristics of the microwave circuit components have been studied. The
microwave power source showed a varying power output level at difi‘erent frequencies.
The microwave power output tended to be higher at higher frequencies. The time for the
computer to write the frequency to the oscillator was measured at about 0.1 seconds. The
cut-off frequency curves measured for the empty cylindrical cavity agreed well with
theoretical predictions, which ensured that the quality of the single mode resonant cavity.

Characterization program for loaded microwave cavity was also designed.
Microwave power reflectance was measured versus frequency. Mode fi'equencies were
obtained by locating the fiequencies with minimal power reflectance. The temperature

profiles measured during the frequency scan were used to determine the variety of heating

147

preferences of the modes, which was used as the criterion for the optimal sample loading
position.

The variable attenuator was characterized to determine the relationship between
the control voltage and attenuation. The relationship turned out to be linear over a wide
range and nonlinear when the control voltage is close to 0. The characterization of the
power meters showed that a certain amount of time was necessary for the meter to give
accurate readings after power step changes. It was determined that the power reading
would be within 5% of the true value after 170 ms of a power step change.

Dielectric properties of DGEBA/DDS showed little change over the fiequency
range from 2 GHz to 4 GHz. The scattered data did not indicate any trends. Therefore, the

frequency effects on epoxy dielectric properties were considered as minimal.

7.2 Variable Frequency Mode Sweeping Heating

The experiments demonstrated that the power source consisting of an oscillator, a
RF plug-in and a TWT worked well as a variable frequency power supply. This variable
fi'equency microwave system, using a cylindrical cavity as the applicator, was able to
obtained heating modes with a variety of heating patterns when graphite/epoxy material
was loaded. A more uniform heating pattern resulted when two modes with
complementary heating preferences were used alternatively for heating.

A mode sweeping heating technique was designed to take advantage of
complementary heating concept and improve the temperature uniformity of microwave

heating. Mode selection cycles were designed. In each mode selection cycle, a sequence of

148

modes was used to heat the sample, each mode for a certain period of time. The mode
selection cycle was repeated until the end of processing.

Mode sweeping method was shown to heat the 2" square graphite/epoxy
composite parts not only uniformly, but also efficiently. The efficiency of mode sweeping
heating lied between the highest one and the lowest one among those of the modes used.
As observed in the experiments, mode sweeping with equal time intervals did not obtain
optimum heating. To get even more uniform heating, the heating characteristics of each
mode should be considered in the optimum heating process design. The heating uniformity
of this method shows the potential to achieve uniform curing of the graphite/epoxy

material of small size.

7.3 Variable Frequency Mode Switching Processing

In variable frequency mode switching heating, processing cycles were designed
based on the mode heating patterns only. However, during the processing of composites,
the temperature distribution varies. Different temperature distributions call for mode with
different heating preferences. An intelligent variable frequency mode switching technique
was designed and developed in LabVIEW (VFMPCSI) to match mode characteristics with
sample temperature distribution when selecting the heating mode.

The control system predicts the temperatures using the mode heating rates and the
measured temperature distribution. The mode that will result in the smallest standard
deviation of the predicted temperatures is selected for heating. A stepper motor was used
to adjust the manual knob on the microwave power source. A parabolic power control

algorithm was designed for simple microwave power control that did not require

149

controller tuning. The power control algorithm was proved to be effective and stable.
However fluctuations occurred during mode switching heating due to the different mode
characteristics and the coarse and slow adjustment of the stepper motor for microwave
power control. 24-ply unidirectional 3" by 3" graphite/epoxy composite parts were
successfiilly processed using the intelligent variable frequency mode switching technique.
Heating experiments proved that the heating characteristics of each mode was repetitive as
long as the sample was loaded in the same way. Mode switching heating resulted in much
improved temperature uniformity.

Variable fi'equency mode switching technique was also employed to significantly
improve the heating uniformity of microwave processing of V-shaped graphite/epoxy
panels. A different microwave power control hardware - variable attenuator, was used.
The power adjustment time for the variable attenuator was much smaller than that of the
stepper motor. The adjustment time using variable attenuator was less than 1 second,
while the adjusment time for stepper motor ranged from 1 second to more than 10
seconds depending on the power change size. The procedure used in this study can be

readily applied to processing of other complex-shaped composite parts.

7.4 Variable Frequency Microwave Processing of Complex Shape Composite
Parts with On-line Mode Updating

An on-line mode characterization capability was added to the variable frequency
microwave processing system and resulted in a much improved process control system -
VFMPCS II. On-line mode characterization was necessary because mode heating

characteristics change during the processing due to material property changes and

150

temperature change. This feature enables the control system to obtain accurate mode
heating characteristics and ensure the effectiveness of the mode selection controller. On-
line mode characterization is especially important for complexly shaped composite parts
since the processing conditions vary in a larger degree than simply shaped composite
parts.

A mode selection controller with a series of prediction steps was designed. Instead
of comparing the heating uniformity at some point in the future, the controller predicts the
temperatures at a number of points in the time after. Only by computing the predicted
temperatures at different points of time, can the potential of each mode to alleviate the
temperature gradient be accurately predicted. A multi-staged power controller was
designed to take advantage of the quick response of the variable attenuator. The
processing of composite materials was divided into four stages, and power control
strategies were designed accordingly. At the heating-up stage, maximum microwave
power was used. When the maximum measured temperature was within 10 °C of the
curing temperature - 160 °C, a PID controller was used for microwave power control.
When the maximum measured temperature was within 160 °C i 1 °C, the second PID
controller was used for microwave power control. The parameters of the two PID
controllers were tuned in such a way that the first PID controller had faster response,
while the second controller was more stable and accurate.

Experiments were conducted for both V-shaped and tri-planar graphite/epoxy
composite parts. Results showed significant improvement of heating uniformity over single
mode heating and mode sweeping heating. The final maximum temperature difi‘erences for

both samples were less than 15 °C. Quantitative comparisons were made for these three

151

processing techniques. For the V-shaped samples, the intelligent mode switching heating
reduced the standard deviation of temperatures by 56% and 75% as compared with mode
sweeping heating and single mode heating respectively. For the tri-planar samples, the
intelligent mode switching heating reduced the standard deviation of temperatures by 26%
and 52% as compared with mode sweeping heating and single mode heating respectively.
The mode selection controller and the on-line mode characteristics updating controller
proved to be effective and accurate, and improved the robustness of the process control
system. The power controller performed well and provided quick, stable, and accurate
curing temperature control. The temperature overshoot was less than 1 °C and the

maximum measured temperature was controlled within 160 °C :1: 0. 5 °C.

7.5 Summary

Compared with previous research results in single microwave cavity processing,
the results accomplished in this work showed significant improvement. Adegbite et al. [20]
used fixed frequency mode switching technique to process 3" 24-ply square
graphite/epoxy composite parts. The final maximum temperature difference was less than
15 °C. However, large temperature fluctuations were present not only in the heating-up
stage, but also in the curing stage. The temperature fluctuation reached 25 °C at times.
Fellows et al. [17] used fixed frequency mode switching technique to process V-shaped
polyester/glass composite parts. The desired curing temperature was 120 °C. The
maximum temperature difference was about 25 °C throughout the processing experiment.
Large temperature fluctuations were also observed and exceeded 15 °C at many occasions.

The curing temperature varied from 120 °C to 130 °C. The reasons for instability of

152

temperature control, large temperature fluctuations, and non-uniform temperature
distribution were not only the slow mechanical adjustment of cavity length for mode
switching, but also lack of a stable and accurate process control system. As comparison,
the results in this work showed uniform temperature distribution throughout the
processing, with maximum measured temperature difference less than 15 °C most of the
time. The temperature control was stable and accurate. No large temperature fluctuations
were observed. Curing temperature overshoot was kept within 1 °C and curing
temperature was controlled within 1 °C of the desired value.

The significance of this work is in the development of a variable fiequency
microwave processing technology that provide uniform and stable processing with
consistent performance and great flexibility and applicability. Advantages of using variable
frequency microwave technology have been explored and demonstrated. A systematic
processing procedure was established, including selection of sample loading positions,
location of the mode fi'equencies, characterization of the heating modes, and finally
computer controlled variable fiequency microwave processing of the materials. A
complete set of variable fi'equency techniques has been created to optimize microwave
processing. The process control system that included optimal mode selection and robust
temperature control has be designed and developed. Specifically, this work made the
following contributions to the microwave material processing technology advancement:

1. The design and implementation of hardware and software for the automation

of a variable fiequency microwave processing system to achieve fast and

precise control.

153

10.

The development and implementation of a process control system using
innovative control methodologies, to achieve uniform and controlled heating
by mode sweeping or switching, mode tuning, and power control.

A microwave cavity characterization program that would determine the
frequencies of the heating modes and the optimal loading position of the
samples.

A predictive mode selection algorithm that would select an optimal heating
mode to alleviate the temperature gradients by matching the sample
temperature distribution with the heating characteristics of the modes.
Power control execution programs that provided fast and precise tuning of the
power control devices, stepper motor and variable attenuator.

An on-line mode heating preference characterization program that would
update the mode heating characteristics database so as to improve the
robustness of temperature uniformity control.

A variable frequency mode tuning program that provides fast and timely tuning
of the mode frequency so as to minimize reflected microwave power.
Analysis and characterization of the performance of microwave circuit
components, such as power meters, in variable frequency processing.
Automatic data acquisition for fast, reliable and convenient data collection,
tracking, and maintenance.

Demonstration of the ability of the variable frequency microwave processing
system to provide uniform and controlled processing of complex-shaped

graphite/epoxy composite parts.

154

11. A tested procedure for variable frequency microwave processing of polymer
composites, including: optimization of sample loading position, location and
characterization of the modes, mode sweeping heating or intelligent mode
switching heating with the option of on-line mode characterization.

12. An intuitive graphical user interface for the operation and control of the

variable frequency microwave processing system.

155

CHAPTER 8

RECOMMENDATIONS AND FUTURE WORK

In this work, variable fi'equency technology has been successfiilly applied in
microwave processing of polymer composites in a single mode cavity. The energy
efficiency of the single mode resonant cavity was exploited while a uniform processing
temperature was achieved, which had been the obstacle of the application of microwave
processing systems in the industry. The advantages of variable fi'equency technology
included more available heating modes, fast mode switching, and easy characterization of

the loaded microwave cavity.

The benefits of variable frequency technology can be firrther realized in on-line
cure monitoring of microwave processing of polymers and composites, because of the
capability of power reflectance scan. Combining thermal heating with variable frequency
microwave heating can fiirther increase the uniformity of the material processing
temperature. As indicated by the experimental results, the temperature gradients at the
curing stage of variable frequency microwave processing were mainly due to the heat loss
from the sample to the ambient. With the utilization of variable frequency technology and
the development of the robust process control system, the microwave processing system
has become a viable system for industrial use. Scale-up studies need to be carried out to

further reduce the gap between the lab-scale system and the industrial processing system.

156

Three recommendations for future work are pr0posed in this chapter. They are: 1)
on-line cure monitoring, 2) hybrid heating for ultimately uniform processing, and 3) scale-
up studies and application of variable fi'equency microwave processing in industrial

processes. These recommendations are presented in the following sections respectively.

8.1 On-line Cure Monitoring for Microwave Processing of Polymers and

Composites

When the composite material is loaded inside a single mode cavity, a power
absorption curve can be obtained by sweeping the fiequency and measuring the percentage
of reflected power. There will be numerous troughs, some of which indicate low
percentage of reflected power, which are characteristics of resonant modes. The
absorption of electromagnetic energy depends on both the fiequency and the material
properties, if other parameters are unchanged. Therefore, if we fix frequency by selecting a
clearly defined trough, the frequency at the dip and the bandwidth will only depend on the
material properties, namely dielectric properties. For the material being processed, its
dielectric properties change with temperature and extent of cure, or chemical composition.
During curing stage, it is desired that the temperature will remain constant, which can be
achieved by the control system. Therefore, by monitoring the change of frequency and
bandwidth of the selected trough, it is possible to detect the change of extent of cure.

In order to accomplish this task, experiments should be carried out to establish the
relationship between extent of cure and dielectric properties of the polymer or composite
material to be processed. An example is given in Figure 8.1. In addition, a theoretical or

empirical model for electromagnetic absorption peak shifi due to dielectric property

157

changes should be developed. Small perturbation method can be used by considering the

change of dielectric properties as small perturbations.

8.2 , Hybrid Heating for Ultimately Uniform Processing

As indicated by the processing results in Chapters 4 to 6, at the curing stage of
variable frequency microwave processing, temperature gradients always existed. The
center temperatures were higher than the rest, especially compared to the temperatures
around the edges of the sample. The cause of the temperature gradients from the center of
the sample to the edges was the heat loss from the sample to the surroundings, because
the ambient temperature was always lower than the sample temperature. The temperature
gradients reduced as the processing progressed and the ambient temperature increased.

For applications that require ultimately uniform temperature distribution during
processing, a hybrid heating approach can be taken by combining variable frequency
microwave heating with thermal heating. Hardware modifications are required for the
microwave processing system. A thermal heater, such as an electrical resistive heating
tape, can be installed around the cavity wall. Computer control of the thermal heater is
crucial to have optimally coordinated heating between microwave heating device and
thermal heating device. A control algorithm will be necessary to control the thermal heater

such that the ambient temperature around the sample will follow the sample temperature.

158

 

AEUV .35:on
4 mm on «m an m 3

1— _ _ — n —

ed vN N.“ N

 

 

 

 

 

 

 

 

Figure 8.1 An Example of Relating Power Absorption Curve Change to Extent of Cure

Change

159

8.3 Scale-up Studies and Industrial Application of Variable Frequency

Microwave Processing System

The variable frequency microwave processing techniques developed on batch
process can be applied to microwave pultrusion process and microwave resin transfer
molding process. Process modeling should be carried out for both pultrusion and Resin
Transfer processes. Heating modes should be identified and characterized. The control
parameters should be determined both empirically and with the aid of mathematical
modeling. Experiments should be conducted to process composite parts using these
systems. The research findings will be the basis of firrther efforts to develop prototype
industrial pultrusion and Resin Transfer Molding systems.

For the scale-up studies, an 18-inch cavity can be used to carry out experiments at
the large scale. A new variable frequency power source with frequency range fi'om 0.5
GHz to 2 GHz will be necessary to provide microwaves than can establish single resonant
modes inside the 18-inch cavity. Proportionally enlarged samples should be processed and
results should be compared with those for 7-inch cavity. Composite parts with complex
geometry similar to industrial products can also be processed so as to investigate the

feasibility of commercializing this technology.

160

APPENDICES

161

APPENDIX A

Control Hardware Instrumentation

1. Stepper Motor Power Control Module
Connecting the data acquisition board with the stepper motor is a 15 pin DSUB connector

for the directional motor control with automatic directional shutoff at end stalls. The

assignments for the pin connections are listed in Table A]

Table A.l Connector and cable wire assignments

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Pin # Wire Color Signal Comments
1 Red 5 V DC External Source
2 - - -
3 - - -
4 - - -
5 Yellow High end TTL Signal -
6 - - -
7 - - -
8 Blue Decrease TTL 3 mA
9 Brown Ground + 5 V DC Return
10 - - -
11 - - -
12 Orange Low End TTL Signal -
13 - - -
14 - - -
15 Green Increase TTL 3 mA

 

 

162

 

2. ARRA 4 752-60D Voltage Controlled Variable Attenuator
The configuration of the ARRA 4752-60D Voltage Controlled Variable is given in Figure

A.l.

Control Terminal V+ terminal

/—— v- terminal

, W
W... , fl 111111

EB

 

Output

EB EB

input _\ 69

 

 

 

Figure A.l Device Configuration of the Variable Attenuator

The general specifications of the variable attenuator are as follows:

Frequency Range ........................ . ......................................... 1.0 _ 18.0 GHZ
Attenuation Range ,, . ~ 1 0 - 60dB
RF Power Max ..... . ..... . .......................................................... +20 dBm

+30 dBm survival

Rise & Fall Time ................................................ , ............... 15 [1 sec] 50 ns
Power Supply .............................. . ............... . ......................... i 12 VOHS, 100 mA
Control Voltage ..................................................... .......... O _ 6 Volts

163

3. Data Acquisition Board - National Instruments” PCI-M10-16XE-50

The National InstrumentsTM PCI-MIO-l6XE-50 (20 kS/s, 16-Bit, 16 Analog
Inputs) is one of the PCI E series boards supplied by National Instruments Inc. This model
of data acquisition boards has bus master capability that makes possible robust,
multitasked DAQ applications. Bus Mastering improves overall system performance
through direct transfer of data between the plug-in board and computer memory, without
burdening the CPU. The PCI standard enables the same application to run on a variety of
operating systems and computers.

The PCI-MIO-16XE-SO board is a multifiinction analog, digital, and timing I/O
board for PCI bus computers. It features l6-bit ADCs with 16 analog inputs, 16-bit DACs
with voltage outputs, eight lines of TTL-compatible digital I/O, and two 24-bit
counter/timers for timing I/O. Because the PCI board has no DIP switches, jumpers, or
potentiometers, it is easily software-configured and calibrated. This feature is made
possible by the National Instruments MITE bus interface chip that connects the board to
the PCI I/O bus. The MITE implements the PCI Local Bus Specification so that the
interrupts and base memory addresses are all soflware configured. Data-acquisition-
related configuration includes such settings as analog input polarity and range, analog

input mode, and others.

Calibration for the PCI-MIO-16XE-50 board:
Calibration refers to the process of minimizing measurement and output voltage
errors by making small circuit adjustments. On the PCI E series boards, these adjustments

take the form of writing values to onboard calibration DACs (CalDACs). Some form of

164

board calibration is required for all but the most forgiving applications. Ifthe board was
not calibrated the signals and measurements could have very large offsets, gain, and
linearity errors. Since both power measurement and power control requires high precision
in this study, calibration was carefully carried out.

There are three levels of calibration available for the PCI E series boards.

1. Loading calibration constants. Loading calibration constants refers to the
process of loading the CalDACs with the values stored in the EFPROM, the onboard
nonvolatile memory. NI-DAQ software dtermines when this is necessary and does it
automatically.

2. Self-Calibration. The PCI board can measure and correct for almost all of its
calibration-related errors without any external connections. The national Instruments
software provides a self-calibration method, which can be initiated by the user. This self-
calibration process, which generally takes less than a minute, is the preferred method of
assuring accuracy. Self-calibration should be sufficient if the user is interested primarily in
relative measurements. Otherwise, the external calibration should be used to address the
gain error due to time or temperature drift of the onboard voltage reference, which could
not be eliminated by the self—calibration process.

3. External Calibration. The PCI E series board has an onboard calibration
reference to ensure the accuracy of self-calibration. This voltage is stable enough for most
applications, but if the board is used at an extreme temperature or if the onboard reference
has not been measured for a year or more, the board needs to be externally calibrated. An

external calibration refers to calibrating the board with a known external reference rather

165

than relying on the onboard reference. The external calibration can be conducted by calling

the NT-DAQ calibration firnction.

Ix’O Connector Pin Assignments:

The pin assignments are illustrated in Figure A.2 [99]. A NationallnstrumentsTM
R6850 Ribbon Cable was used to connect the 68-pin data acquisition board to a 50-pin
I/O connector block. This greatly simplified the labeling associated with the pins. The U0

connector block terminals are listed in Table A.2 along with corresponding signals.

166

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ACH8 - 34 68 - ACHO
ACHl- 33 67 -AIGND
AIGND- 32 66 -ACH9
ACHIO- 31 65 -ACH2
ACH3- 3o 64 -AIGND
AIGND- 29 63 -ACH11
ACH4- 28 62 -AISENSE
AIGND- 27 61 -ACH12
ACH13- 26 60 -ACH5
ACH6- 25 59 -AIGND
AIGND- 24 58 -ACH14
ACH15- 23 57 -ACH7
DACOOUT- 22 56 -AIGND
DACIOUT- 21 55 -AOGND
UNA SSIGNED - 20 54 - AOGND
9104- 19 53 -DGND
DGND- 18 52 -D100
D101- 17 51 -DIOS
DIO6- 16 5o -DGND
DGND- 15 49 -0102
+5v- 14 48 -DIO7
DGND- 13 47 -DIO3
DGND- 12 46 -SCANCLK
PFIO/TRIGI - 11 45 -EXTSTROBE
PFIl/TRIGZ- 10 44 -DGND
DGND - 9 43 - PFIZ/CONVER'I‘
+5v- 8 42 -PF13/GPCTR1_SOURCE
DGND- 7 41 -PFl4/GPCTR1_GATE
PFIS/UPDATE - 6 40 -GPCTR1_OUT
PFl6/WFI'RIG- 5 39 -DGND
DGND - 4 38 - PFl‘I/STARTSCAN
PFl9/GPCTRO_GATE - 3 37 - PFl8/GPCTRO_SOURCE
GPC’I'R0_OUT - 2 36 - DGND
FREQOUT- 1 35 -DGND

 

Figure A.2 Pin Assignments for PCI-MIO-léXE-SO Board

167

Table A.2 I/O Connector Block Terminals and Corresponding Signals

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PHYSICAL PIN ASSIGNMENTS PIN # PIN ASSIGNMENTS PHYSICAL
SIGNAL SIGNAL
AIGND 1 2 AIGND
Cavity Length ACHO 3 4 ACI-I8 Temperature 1
Probe Depth ACI-Il 5 6 ACH9 Temperature 2
Input Power ACH2 7 8 ACHIO Temperature 3
Reflected Powep ACH3 9 10 ACHll Temperature 4
ACH4 11 12 ACH12 Tempepgture 5 ...,-
ACHS l3 l4 ACH13 Temperature 6
- .. i ACH6 15 16 ACH14 Temperature 7
ACH7 17 18 ACHIS Temperature 8
AISENSE 19 20 DACOOUT VA_V/SM_Inc]
SM_Dec2 DAClOUT 21 22 UNA SSIGNED
AOGND 23 24 DGND
Probe Direction D100 25 26 D104
Probe Pulse 0101 27 28 D105
Cavity Direction D102 29 30 D106
Cavity Pulse 0103 31 32 D107
DGND 33 34 + 5 V Probe/SM Power
Cavity Power + 5 V 35 36 SCANCLK
EXTSTROBE 37 38 PFIO/TRIGI
PFIl/TRIG2 39 40 PFIZ/CONVERT J
PFI3/GPCTR1_SOUR 41 42 PFI4/GPCTR1_GAT
3 CE E
GPCTR1_OUT 43 44 PFIS/UPDATE
PFI6/WPTRIG 45 46 PFI7/STARTSCAN
PFI8/GPCTRO_SOUR 47 48 IPFI9/GPCTRO_GAT
CE E
GPCTRO_OUT 49 50 FREQ_OUT
Notes:

1. Pin #20 was connected to either VA_V or SM_Inc, where VA_V is the control voltage

for the variable attenuator and SM_Inc is the " increase" voltage for the stepper motor.

2. SM_Dcc is the "decrease" voltage for the stepper motor.

168

4. Schematic for Teflon Molds

All Teflon molds were composed of two parts, a top part (cover) and a bottom part
(holder). Latches are used to put the two parts together. The Diameters for the latch holes
are all 0.375 ". Eight probing holes are to be drilled with locations and dimensions

sketched below. The diameters for the probing holes are all 0.125 ”.

1. Teflon Mold for V-Shaped Samples

A Teflon mold was made for a V-shaped composite part. The mold is consisted of a
holder and a cover. The mold looks like a cylindrical block when closed. The sample is
loaded in the center of the mold. The schematics for the Tri-planar Teflon mold are

presented in Figure A.6 through Figure A.8.

 
  
   
 

 

fl

3.375":

\1 1.625" 1.625"

 

 

 

 

 

 

(a) Top View

169

—1——
0375"},

 

70.375"

 

 

Lat—v

 

.4 1w

 

 

H++

 

 

 

 

 

 

(b) Side Views

Figure A.3 Schematic for V-Shaped Teflon Mold Cover

 

   

1.625"

 

 

 

 

 

 

 

 

 

(a) Top View

 

 

....~ .......
...........

A

1.5"

 

 

I

4+4»

«’4»

 

 

 

 

 

 

 

Figure A.4 Schematic for V-Shaped Teflon Mold Holder

170

(b) Side Views

0.375"

  

 

< 0.25"
1

D: 0.125"

 

 

 

 

 

 

 

 

 

/
/ \
A
{I (x t‘ a
it K2 jx K? \
A 5" I .0 \nf . I"
:0" 0 <5 V ..
t 0.5" t: f,“ j I I"
' xx xx rx I
l VJ J ,7
‘ I

 

 

 

 

 

 

 

 

(a)

0.375"

Cover

 

 

 

 

(b) Holder

Figure A.5 Schematic for V-Shaped Teflon Mold Latches and Probing Holes

171

 

2. Teflon Mold for Tri-Planar Samples

A Teflon mold was also made for processing tri-planar composite parts. The mold is
consisted of a holder and a cover. The mold looks like a cylindrical block when closed.
The sample is to be centered in the mold. The schematics for the Tri-planar Teflon mold

are presented in Figure A.6 through Figure A.8.

 

 

1.125” 1.125"

r ‘7

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(a) Top View
L t
0 28m A / l"
11 1 L 0.25" rur {l L
W TT
(1)) Side Views

Figure A.6 Schematic for Tri-Planar Teflon Mold Cover

172

 

   
 

 

3.375"

  

l.l25"

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(a) Top View
44 A I
II ......................... f) S n 7
+ .. I '~ H
x“... ...,...11. ................ - 1 -5 it ‘
(b) Side Views

Figure A.7 Schematic for Tri-Planar Teflon Mold Holder

173

0.375"

   
   
   

0.25"

D: 0.125"

0.875 0.875

 

 

 

 

(a) Cover
0 .3 7 5 "
\ 0 2 5 "
I -~~\“
I" "\_
o'i/ \\
/ x.
If
I
‘t
\. 4"
'\. ,’
‘\. .v/
\‘.\ '/
\.\' I"

 

(b) Holder

Figure A.8 Schematic for Tri-Planar Teflon Mold Latches and Probing Holes

174

APPENDIX B

LabVIEW Subvi's

1. f-write#. vi

This program sets single microwave frequency. A formatted string for the
frequency is written to the Oscillator through the GPIB interface. “CW” sets the mode of
operation for the oscillator, which is followed by frequency value and unit. The GPIB
address of the oscillator is 19.

The LabVIEW program front panel and diagram of f-write.vi are presented in
Figure B. 1.
2. valstep. vi

This program changes the attenuation of the variable attenuator by changing the
control voltage. The control voltage of the variable attenuator can be controlled on the
front panel. The device number for the National Instrument Data Acquisition board is 1.
The attenuator control voltage is connected to the analog output 0 of the DAQ board
(Figure A.2). The control voltage range is O — 6 volts.

The LabVIEW program fiont panel and diagram of va—l-step.vi are presented in
Figure B.2.
3. vapwrctrl. vi

This program is used for variable attenuator (microwave) power control. Given a
desired microwave power, this program adjusts the control voltage so that the microwave

power output is close enough to the desired power. The logarithmic relation between the

175

control voltage and the attenuation was used to provide a starting estimate of the control
voltage. A linear search method based on the logarithmic relation was used for tuning of
the control voltage. The measurement of microwave power is carried out after 175 ms of
each control voltage change.

The LabVIEW program front panel and diagram of vactrl.vi are presented in
Figures B.3 and B.4.
4. pwrctrl.vi

This program is used for the control of microwave power using stepper motor.
The desired microwave power is compared with the measured microwave power. If the
desired power is higher than the measured power, the polarity of the control voltages are
such that the stepper motor turns in the increasing direction of the microwave power, and
vice versa. At each stepper motor adjustment, the control voltages are applied for 50 ms
and then zero voltages are applied to halt the stepper motor for 150 ms. In this way, more
stable stepper motor adjustment is achieved.

The LabVIEW program front panel and diagram of pwrctrl.vi are presented in
Figures 3.5 and B.6 respectively.
5. m-tuning.vi

The purpose of this program is used for mode tuning. The frequency is tuned
within the given range in order to minimize the reflected microwave power. The
granularity of frequency tuning can be adjusted.

The LabVIEW program front panel and diagram of mtuningvi are presented in

Figures B.7 and B.8 respectively.

176

 

.Efiffi: fllbtmlogl-‘ont

 

 

 

Figure 8.1 LabVIEW Program of f-write#.vi - Front Panel and Diagram

177

 

 

    

_g- sttouldboldrr

 

Figure B.2 LabVIEW Program of valstep.vi - Front Panel and Diagram

178

 

V I ll 111

f3. IEIWIPFSEl—lmlm

 

 

 

 

Figure B.3 LabVIEW Program of vapwrctrl.vi - Front Panel and Diagram

179

 

 

 

 

 

Figure B.4 Additional Elements of vapwrctrl.vi ~ Diagram

I80

 

Figure 3.5 LabVIEW Pogram of pwrctrl.vi - Front Panel and Diagram

181

Flue

.. ~ 1’8deth

 

 

 

 

 

 

Figure B.6 Additional Elements of pwrctrl.vi Diagram

182

 

Figure B.7 LabVIEW Program for m-tuning.vi - Front Panel and Diagram

183

 

 

Figure B.8 Additional Elements of m-tuning.vi Diagram

184

APPENDIX C

LabVIEW Programs for System Characterization

1. vapwrtest. vi (Characterization of Variable Attenuator)

This program is used to characterize the variable attenuator in order to determine
how the relationship between attenuation and control voltage changes with frequency.
Microwave power was measured while changing the control voltage at different
frequencies.

The LabVIEW program front panel and diagram of vapwrtest.vi are presented in
Figures Cl and C2 respectively.

2. p-response-test. vi (Measurement of Power Meter Response Time)

The response time of the power meters was measured using this program.
Microwave power is measured as firnction of time after a power step change due to
variable attenuator control voltage change. The magnitude of power change can be varied.

The LabVIEW program front panel and diagram of p-response—test.vi are
presented in Figures C.3, C4, and C5 respectively.

2. Characterization& temp. vi (Measurement of Power Reflectance Curve and
Temperature Change)

Using this LabVIEW program, the power reflectance curve is measured, along
with the temperature change while varying the frequency. Six temperatures are measured,
which can be expanded to eight. During the run, frequency is changed from lower end to

higher end with specified increment. The input and reflected microwave powers are

185

measured and the reflectance is computed and plotted. Temperatures are measured during
the fi'equency sweep.
The LabVIEW program front panel and diagram of characterization&tempe.vi are

presented in Figures C6, C7, C8, and C.9 respectively.

186

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure C.l LabVIEW Program of vapwrtest.vi - Front Panel and Diagram

187

   
  

IIIIII l ‘l'li’Gl‘ll l 1 I614"-

 

 

 

 

 

Figure C.2 Additional Elements of vapwrtest.vi Diagram

188

 

 

Figure C.3 LabVIEW Program of p-response-test.vi - Front Panel

 

Figure C.4 LabVIEW Program of p-reponse-test.vi - Diagram

189

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure C.5 Additional Elements of p-reponse-test.vi Diagram

190

 

 

 

 

 

 

llZptDilquoM

pg‘":
1505‘ I

v.

.1
L1

l
. 1
.
1 i
i .
.‘1 i
l ,
lY ‘ i
Y .
I
, .
. .
1: ,
411,
11
Ml
11
1‘1
7‘ ‘
,1 1
D ‘
’11 »
[1‘1 1‘
‘l
Y . .
, .
l1 .
'.1
1' .
.
:1"
I I
I" 4 ‘
‘ l
l
hi
‘ . 1 .
1 I l
‘12
1.
pl
*1 u -
fir»:
L11
i I

Figure C.6 LabVIEW Program of '

191

 

 

p...__: vi - Front Panel (Lefi Half)

 

 

(Right Half)

FIGURE C.7 LabVIEW Program of characterization&temp.vi - Front Panel

192

 

 

 

 

Figure C.8 LabVIEW Program of characterization&temp.vi - Diagram

193

 

 

3+1

 

 

 

L..." "__._—*1

.. BMNRJW ]
I
i
1

. __._—J

 

 

 

Figure C.9 Additional Elements of characterization&temp.vi Diagram

194

 

 

 

 

 

 

[31

 

 

 

 

Figure C.9 (Continued)

195

 

APPENDIX D

LabVIEW Programs for Process Control System

1. Singlemode. vi - single mode heating

This is the LabVIEW program for microwave heating experiment using a single
mode with PID microwave power control. The heating experiment starts at room
temperature with the initial frequency. The program shown measures six temperatures at
the sample surface. As the temperatures approach the curing temperature, a PID control
algorithm is used to maintain the highest temperature at the curing temperature. The user
can determine how long the experiment lasts.

As the sample is being heated, the program tunes the fi'equency in the vicinity of
the initial frequency in order to minimize the reflected microwave power. The tuning
ranges and tuning granularity can be different for heating stage and curing stage. A tuning
period is used to determine how often the frequency is tuned. A minimal microwave
power is required for the frequency tuning to ensure the accuracy.

Other parameters that can be adjusted include power measurement and control
parameters and temperature measurement and control parameters. Two sets of PD)
control parameters are used for the two-staged PID control. The microwave power
reflectance curve and the temperature profiles are displayed on the LabVIEW control
panel

The LabVIEW program front panel and diagram of singlemode.vi are presented in

Figures D. 1, D2, and D3 respectively.

196

2. modesweep. vi - variable frequency mode sweeping heating

Six modes are used in this program for the microwave heating experiment. These
modes are used in a sequence with each mode assigned a heating time. Frequency tuning is
used to minimize the reflected microwave power for each mode. A two-staged PID
control algorithm is used for the microwave power control. The sample is heated from
room temperature and the maximum temperature is maintained at the curing temperature
by PID control. An on-line monitoring option is provided for measuring the minimum
reflectance frequency for a selected mode. The user can adjust the parameters for PID
control, on-line characterization, mode tuning, and power and temperature measurement.
Temperatures are plotted on the control panel.

The LabVIEW program front panel and diagram of modesweep.vi are presented in

Figures D.4, D.5, and D6 respectively.

2. VFMPCSI. vi - variable frequency mode switching heating with process control
This program is used to process composite parts following the similar procedure as
described in singlemode.vi . The number of measured and controlled temperatures can be
changed in the program. The major difference fi'om singlemode.vi is that in this program,
mode switching technique is implemented to achieve more uniform heating. In addition,
the power control algorithm uses a parabolic relation between the temperature and the
microwave power. The mode switching algorithm is described in details in Section 5.2.
The LabVIEW program front panel and diagram of vfmpcsI.vi are presented in

Figures D.7, D8, and D.9 respectively.

197

3. VFMPCSII. vi - variable frequency mode switching heating with process control
and on-Iine mode characteristics updating

This program is based on VFMPCSI.vi and uses a two-stage PID control
algorithm. The mode switching algorithm follows a similar idea but uses a different
implementation compared with that of VFMPCSI.vi. Furthermore, this program adopts an
On-line Mode Characteristics Updating Controller to adapt the mode selection to the
heating characteristics change of the modes. The details of the algorithms are presented in
Section 6.2.

The LabVIEW program fiont panel and diagram of vfmpcsII.vi are presented in

Figures D. 10, D. 1 1, and D.12 respectively.

198

 

d..81.l;l ill Ill-tilt it .1

k
W
W

m

 

 

eaten—.5355
griigkgwom‘ou‘oa

W
W
m
M
L
V
.

 

Front Panel

Figure 0.1 LabVIEW Program of singlemode.vi

199

 

Figure D.2 LabVIEW Program of singlemode.vi - Diagram

200 ‘

 

 

 

 

,1

 

.1. VA
E ‘3th

them-aw-

 

INo mode tuning if there's no mode change I

 

Ino on-line monitoring data I

 

 

 

 

Figure D.3 Additional Elements of singlemode.vi Diagram

201

 

 

eb-

 

 

Front Panel

Figure D.4 LabVIEW Program of modesweep.vi -

202

 

Figure 0.5 LabVIEW Program of modesweep.vi - Diagram
203

 

 

      

 

1£If§=lfll .I. hm}..— ,1 lug- H
I -' udesrvertoleve'.
. A9 :1:

».‘JlLIti' I

 

 
   
       

 

Ifilelulufl l Irin-

 

 

   
        

 

 

 

      
 
 

 

 

 

 

 

 

 

 

 

 

 

Case 2
— ‘1 L Case 3
ralse 2 ,
L21 .1 False 1
no mode tumng.
two cases mode tuning activates:
1. mode change;
2. mode has been heating for
a cerain riod of time.
EIralseEl
Ino on-Iine monitoring data I
I:
lused as em arra
III-abet]

 

 

 

 

 

 

 

 

 

Figure D.6 Additional Elements of modesweep.vi Diagram

204

v.0-

 

 

 

 

Front Panel

.Vl -

7 LabVIEW Program of VFMPCSI

reD

Fig“

205

 

Figure 0.8 LabVIEW Program of VFMPCSI.vi - Diagram
206

 

N.

7
nl\..

S-uence1

 

 

 

 

 

til-ales LEI

 

 

[__W

 

Figure D.9 Additional Elements of VFMPCSI.vi Diagram

207

 

 

 

C .

 

EIFalse LEI

 

Ino action if (i=power of! Lequency)

 

 

 

 

 

 

 

 

 

 

 

action if actual
is within range

max power (lnttia
- fereour .51 .

 

Figure D.9 (continued)

208

 

,3.

u 1.11-1191!

Front Panel

i-

10 LabVIEW Program of VFMPCSIIN

D

igure

F

209

11411 villi) .1114... it i
. . a .. .

. an.“

 

 

 

 

 

renew ten . . data

:. I>

.

Figure 0.1 l LabVIEW Program of VFMPCSll.vi - Diagram
210

mode uhanged’
Curing Stage
U . date 2

mode changed or updating
or on~llne monitorin-

ED

Curing T
I:> Reached’

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure D.12 Additional Elements of VFMPCSII.vi Diagram

211

 

fiiorcgfigE-ldqrihtooauvoopxeé_ . .
goggouitfifiih e5. buggeé. L

 

 
   

  

 

 

 

 

 

 

 

 

 

 

 

 

 

um 55 2 ouoE can}.
25 3:33.: 3:93

6935 5:252. o:

 

] _fimfit

 

 

 

 

   

Figure D.12 (continued)

212

10.

11.

12.

13.

14.

15.

16.

17.

BIBLIOGRAPHY

White, W. C., Proc. Inst. Radio Electron. Eng., August, p50, 1129 (1962)

Chabinsky, I. J. , Materials Research Society Symposium Proceedings, p124, 17
(1988)

Krieger, B., Proc. Am. Chem. Soc., Div. Polym. Mater. : Sci. Eng, 66, p339 (1992)

Gourdenne, A., A. H. Maassarani, P. Monchaux, and S. Aussudre, Polym. Prepr. ,
Am. Chem. Soc., Div. Polym. Chem, 20(2), p471 (1979)

. Lewis, D., J. C. Hedric, T. C. Ward, and J. E. McGrath, Polym. Prepr., Am. Chem.

Soc., Div. Polym. Chem. 28(2), p330, (1987)

Lewis, D., Mater. Res. Soc. Symp. Proc. , 269, 21 [Microwave Processing of Materials
111] (1992)

low, J., J. Delong, and M. C. Hawley, SAMPE Q., 20, p46 (1989)

Methven, J .M. and S. R. Ghafi‘ariyan, Principles of Microwave Heating (Binner,
J.G.P., ed.), pp.56—77, Abington Publishing

Wei, J ., J. D. Delong, M. DeMuse, and M. C. Hawley, Polym. Eng. Sci, 33, p1132
(1993)

Beldjoudi, N. and A. Gourdenne, Eur. Polym. J., 24, 265 (1988)

George, C. E., G. R. Lightsey, and A.G.Wehr, Materials Research Society Symposium
Proceedings, 124, pl89[Microwave Processing of Materials] (1988)

Agrawal, R. and L. T. Drzal, J. Acfltes., 29, 63 (1989)
Wei, 1., Ph.D. Dissertation, Michigan State University (1992)
Asmussen, J ., H. H.Lin, et al., Rev. Sci. instrum, 1477-1486 (1987)

Chen, Y. F., and C. Y. C. Lee, Polym. Mat. Sci. Eng, Am. Chem. Soc., 60, pp680-
684 (1989)

Volgel, G. L., J. J ow, et al., 4’” Technical conference of Composite Materials, (1989)

Fellows, L. A., R. Delgado, and M. C. Hawley, 26th International SAMPE Technical
Conf, Atlanta, GA, Oct. (1994)

213

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

Wei, J., J. Jow, J. D. Delong, and MC. Hawley, SAMPEJ, 27(1), p.33 (1991)
Wei, J ., and MC. Hawley, [MRS Symposium Proceedings, 269, 439 (1992 )
Adegbite, V. O., Ph.D. Dissertation, Michigan State University (1995)

Harrington, R. F ., T ime-Harmonic Electromagnetic Fields, McGraw-Hill Book
Company (1961)

Chen, Kun-Mu, Electromagnetic Theory 1, Class Notes, Department of Electrical
Engineering, Michigan State University (1995)

Tai, Chen-To, Dyadic Green Functions in Electromagnetic Theory, IEEE Press, New
York, p.47 (1994)

Pozar, D. M., Microwave Engineering, Addison-Wesley Publishing Company, Inc.
(1991)

Collin, R., Foundations for Microwave Engineering, McGraw—Hill (1966 )

Ramo, S. 311d Whinnery, J. R., Fields and Waves in Modern Radio, John Wiley &
Sons, New York (1944)

Ku, CC. and R. Liepins, Electrical Properties of Polymers: Chemical Principles,
Macmillan Publishing Company (1987)

Bruce, R.W., Mat. Res. Soc. Symp. Proc., 124, 3 (1988)
Lee, W. I. and G. S. Springer, J. Compos. Mater., 18(4), p.357 (1984)

Lewis, D. A., T. C. Ward, J. D. Summers, and J. E. McGrath, Polymer Preprints,
29(1), pp.174-175 (1988)

Lewis, D.A., T. C. Ward, J. D. Summers, and J. E. McGrath, Journal of Polymer
Science: Polymer Chemistry, 30(8), pp. 1647-1653 (1992)

Bush, S.F . and J. M. Methven, Proceedings of the 39'” International SAMPE
Symposium, Anaheim, April 11-14, pp.34-43 (1994)

National Materials Advisory Board Commission on Engineering and Technical
Systems, (National Research Council), Microwave Processing of Materials, National
Academy Press (1994)

Hottong, U., J. Wei, R. Dhulipala, and M. C. Hawley, Ceram. T rans., 21, p587 (1991)

Mashida, T., S. Matsushita, and I. Ikegami, Japanese Patent 50 138 067 (1975)

36. Jullien, H. and H. Valot, Polymer 26, 505 (1985)

214

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

Gagliani, J ., R. Lee, and A. L. Wilcoxson, US Patent 4 305 796 (1981)

Lewis, D., Proceedings of the 39’” International SAMPE Symposium and Exhibition,
pp. 1-12, (1994)

DeMuse, MT and A. C. Johnson, Proceedings of the 29‘” Mcirowave power
Symposium, pp. 168-171 (1994)

Thuillier, FM. and H. Jullien, Makromol. Chem. : Macromol. Symp., 25, p63 (1989)
Mijovic, J., A. Fishbain, and J. Wijaya, Macromolecules, 25, p.986 (1992)
J ow, J ., Ph. D. Dissertation, Michigan State University (1988)

Lind, A.C., L. N. Medgyesi-Mitschang, J. E. Kurz, H. F. McKinney, and RC. Wear,
Mater. Res. Soc. Symp. Proc., 189, p.461 [Microwave Processing of Materials 11]
(1991)

Ippen, J ., Rubber Chem. T echnol., 44, p.294 (1971)
VanKoughnett, AL. and J. G. Dunn, J. Microwave Power, 8, p.101 (1973)
Wang, C.S., Rubber Chem. T echnol., 57, p.134 (1984)

Methven, J .M., Proceedings of the 39'” International SAMPE Symposium and
Exhibition, pp. 25-33 (1994)

Wei, J ., B. Thomas, and M. C. Hawley, SPE Proc.: ANTEC ’92 (Vol. 1), p. 1170
(1992)

Lin, M. , and M. C. Hawley, Proceedings of the 38’” International SAMPE Symposium
andExhibition, pp. 436-445 (1993)

Qiu, Y., A. Smith, J. Wei, M. Lin, and MC. Hawley, Composites in Manufacturing,
SME Quarterly, 1't Quarter, (2000)

Falconer-Flint, M.J., Aust. Plast. RubberJ, 24(11), 33 (1973)

Clark, D.E., Ceramic Engineering and Science Proceedings, v14, n7-8, ptl, pp.3-21
Westville, OH (1993)

De’,A.S., I. Ahmad, E. D. Whitney, and D. E. Clark, Mater. Res. Soc. Symp. Proc.,
189, 283-288, [Microwave processing of Materials H] (1991)

Patterson, M. C. L., P. S. Apte’, R. M. Kimber, and R. Roy, UIE Microwaves and
Radio Frequency Conference Proceedings, pp. 33-3 6, Nice, France (1991)

215

55. Patterson, M. C. L., P. S. Apte’, R. M. Kimber, and R. Roy, Mat. Res. Soc. Symp.
Proc. , [Microwave Processing of Materials 111] (1992)

56. Acitally, M. A., R. B. Prime, and E. Sacher, Polymer, pp. 335-342 (1971)

57. Hagnauer, G. L., P. J. Pearce, B. R. LaLiberte, and M. E. Roylance, Ihermorheology
of Thermosetting Polymers, pp. 25-47 (1983)

58. Dutta, A. and M. E. Ryan, Journal of Applied Polymer Science, (24), p.635 (1979)
59. Mijovic, J ., Journal of Applied Polymer Science, (31), p. 1 177 (1986)

60. Moroni, A., J. Mijovic, E. M. Pearce, and C. C. Foun, Journal of Applied Polymer
Science, (32), p.376] (1986)

61. Sichina, W.J., Dupont Company Applications Brief, No. TA-93

62. Gillham, J. K., Developments in Polymer Characterization -3, Dawkins, J .V., p.159,
Applied Science, London (1982)

63. Sourour, S. and M. R. Kamal, Ihermochimic Acta, (14), 41 (1976)

64. Lee, W.I., A. C. Loos, and G. S. Springer, Journal of Composite Material, (16), 510
(1982)

65. Jow, J ., M. C. Hawley, M. F inzel, and T. Kern, Polymer Engineering and Science,
28(22), pp. 1450-1454 (1988)

66. Delmotte, M., H. Jullien, and M. Ollivon, Eur. Polym. J., Vol. 27, No. 4/5, pp.371-
376 (1991)

67. Delmotte, M. and H. Jullien, Makromolek. Chem. ; Macromolec. Symp., 9, 185 (1987)

68. Marand, E., K. R. Baker, and J. D. Graybeal, Radiation Eflects on Polymers,
American Chemical Society, pp.385-400 (1991)

69. Livi, A., G. Levita, and P. A. Rolla, Journal of Applied Polymer Science, Vol. 50,
pp.1583-1590 (1993)

70. Wei, J ., Y. Chang, B. Thomas, and M. C. Hawley, Proc. ICCM/VIII, vol.I, 10-L
Honolulu, Hawaii (1991)

71. Lin, M. and M. C. Hawley, Proceedings of the 38‘” SAMPE International Symposium,
Anaheim, CA (1993)

72. Willams, N.H., Journal of Microwave Power, 2, pp. 123-127 (1967)

73. Gourdenne, A. and V. Q. Le, Eur. Polym. .1., 23(10), pp.777-780 (1987)

216

74.

75.

76.

77.

78.

79.
80.

81.

82.

83.

84.

85.

86.

87.

88.

89.

90.

91.

92.

Fellows, LA. and M. C. Hawley, Proceedings of the 29th Microwave Power
Symposium, p.91, Chicago, IL (1994)

Wei, J. et al., SAMPEJournal, 27 (1), 33 (1991)
Lee, W.I., and GS. Springer, J. Composite Materials, 18(4), 357 (1984)
Kashyap, SC, and W. Wyslouzil, J. Microwave Power, 12(3), pp. 223-230 (1977)

Espinosa, R.J., et al., 28th Microwave Symposium Proceedings, Manassas, VA,
pp.26-31 (1993)

Garard, R.S., Mat. Tech, 9, pp.235-238 (1994)
Collins, R.E., Flow of Fluids Through Porous Materials, Reinhold, New York (1961)

Jow, J ., M. C. Hawley, M. C. Finzel, and J. Asmussen, Rev. Sci. Instrum, 60, 1, p.90
(1989)

Watanabe, M., M. Suzuki, and S. Ohkawa, Journal of Microwave Power, 13(2),
pp.173-181 (1978)

Zhu, S. and H. Chen, Journal of Microwave Power and Electromagnetic Energy,
23(3), pp.139-143 (1988)

Paoloni, B, Journal of Microwave Power and Electromagnetic Energy, 24(1), pp.2 1 -
32 (1 989)

Lin, H., Ph.D. Dissertation, Michigan State University (1989)
Manring, 3., Ph. D. Dissertation, Michigan State University (1992)
de Pourcq, M., [E Proceedings, Pt.H, 131(6), pp.411-419 (1984)

J ia, X. and P. Jolly, Journal of Microwave Power and Electromagnetic Energy,
Vol.27, No.1, pp.11-22 (1992)

Volpe, V., Journal of Composite Materials, 14, p.189 (1980)

Morgan, M.A., D. L. Fisher, and E. A. Milne, IEEE Transaction on Antennas and
Propagation, Ap-3 5(2), p.191 (1987)

Hashimoto, O. and Y. Shimizu, IEEE Transactions on Microwave Theory and
Techniques, MTT-34(11), p. 1202 (1986)

Wei, J ., M. C. Hawley, and J. Asmussen, Journal of Microwave Power and
Electromagnetic Energy, 28(4), p.234 (1993)

217

93.

94.

95.

96.

97.

98.

99.

Hinrichs, R. J. , Composite Materials: Quality Assurance and Processing, ASTM STP
797, pp. 29-37 (1983)

Alliouat, M., Y. Lecluse, J. Massieu, and L. Mazo, Journal of Microwave Power and
Electromagnetic Energy, v25, n1, pp25-31 (1990)

Beale, G. 0., F. J. Arteaga, and W. M. black, IEEE Transactions on Industrial
Electronics, Vol. 39, No. 4, pp. 301-312, August (1992)

Sticklen, J ., A. Kamel, M. C. Hawley, and V. Adegbite, IEEE Expert, pp. 43-53, April
(1992)

Hawkins, R., M. C. Hawley, T. Lenz, J. McDowell, Y. Qiu, and J. Sticklen,
Proceedings of A merican Society of Composites 12‘” Technical Conferences,
Dearborn, MI (1997)

Hawkins, R., M. Hawley, J. Sticklen, Y. Qiu, and J. McDowell, Proceedings of
AICHE Annual Meeting, Chicago, IL (1996)

National Instruments, DAQ - PCI E Series User Manual (1998)

218

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

"1'“1"11111111111111.1111”1'